Intelligent prediction method for intelligent jacking construction progress of bent cap

Abstract

The invention relates to the technical field of intelligent construction, and discloses a bent cap intelligent jacking construction progress intelligent prediction method, which comprises the following steps: collecting multi-source real-time data such as equipment, structure, personnel and environment in a construction process, and calculating a construction entropy for measuring system uncertainty; establishing a causal state evolution model, and predicting the evolution trend of the construction entropy in a future time period; and when it is predicted that a construction progress disturbance risk exists, performing anti-fact deduction based on the model or the dynamically trimmed sparse model thereof, and generating and recommending an optimal adaptive correction strategy. According to the method, the overall uncertainty of the construction system is quantified into the construction entropy, and the prospective deduction capability of the causal state evolution model is combined, so that the conversion from the risk passive response depending on experience to the data-driven beforehand active early warning and optimal decision recommendation is realized; and improvement of scientificity, foresight and timeliness of construction risk management is facilitated.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent construction, in particular to a method for intelligently predicting the construction progress of a bent cap. BACKGROUND

[0002] Bent cap intelligent lifting construction, as a key link of bridge and elevated structure upper structure installation, refers to the construction process of precisely and smoothly lifting the prefabricated or cast-in-place bent cap structure from the ground or low support to the design elevation by using the integrated control hydraulic synchronous lifting system. The process has high sensitivity to equipment synchronization, structure posture and construction environment.

[0003] The existing bent cap lifting construction progress monitoring technology mainly relies on the sensor array arranged at the key equipment and structure parts, acquires and displays the isolated physical quantity parameters such as the pressure, displacement of each jack, and the stress and inclination angle of the key section of the bent cap in real time, and alarms in combination with the preset single parameter safety threshold. The judgment of construction progress and risk state mainly relies on the manual interpretation and comprehensive experience of the on-site total control engineer on the discrete data.

[0004] Since the bent cap lifting construction is a complex dynamic system closely coupled by equipment, structure, personnel and environment, there is a complex nonlinear causal relationship between the monitoring data, which makes it difficult for the existing technology to understand the overall evolution trend of the system state from the massive and multi-source apparent data. Therefore, there are inherent defects such as serious lag in identifying construction risks, lack of forward-looking prediction ability of the future state of the system, and inability to quantitatively evaluate the overall uncertainty level of the system. This defect easily causes the gradual accumulation and amplification of minor construction disturbances due to the failure to be discovered and corrected in time, and eventually leads to serious lag in construction progress, structure damage and even safety accidents and other adverse consequences. SUMMARY

[0005] In view of the deficiencies of the prior art, the present application provides a method for intelligently predicting the construction progress of bent cap intelligent lifting construction, aiming to solve the problem of lack of overall and forward-looking assessment of construction risks in the prior art, leading to lag in risk identification and lack of scientific correction decision basis.

[0006] To achieve the above purpose, the present application is realized by the following technical scheme: A method for intelligently predicting the construction progress of bent cap intelligent lifting construction, comprising the following steps: Collecting multi-source real-time data in the bent cap lifting construction process, and preprocessing the multi-source real-time data to obtain a standardized state data set; Based on the standardized state data set, constructing and calculating construction entropy for measuring the uncertainty of the construction system; establish a causal state evolution model including observable variables, hidden variables and macro state variables, and based on the current construction system state, use the model to deduce and predict the construction entropy evolution trend in the future period; monitor the value and its rate of change of the construction entropy in real time, and when the value or its rate of change of the construction entropy exceeds a preset threshold, dynamically prune the causal state evolution model to generate a sparse model focusing on the core causal path; When the construction entropy evolution trend indicates that there is a disturbance risk in the construction progress, perform counterfactual reasoning based on the causal state evolution model or the sparse model to generate an adaptive correction strategy that can reduce the construction entropy in the future period.

[0007] Preferably, the multi-source real-time data includes: device state data, structure state data, personnel operation data and environmental state data.

[0008] Preferably, the construction entropy is obtained by weighted summation of device state entropy, structure state entropy, personnel state entropy and environmental state entropy through a preset weight coefficient.

[0009] Preferably, the calculation method of any one of the device state entropy, structure state entropy, personnel state entropy and environmental state entropy includes: Within a sliding time window, for the multiple parameters corresponding to the state entropy, calculate the deviation sequence of each parameter relative to its baseline value; Discretize the deviation sequence to obtain multiple discrete state intervals; Statistical state probability of the deviation sequence in the multiple discrete state intervals; Apply the information entropy formula to calculate the value of the state entropy according to the state probability.

[0010] Preferably, the causal state evolution model is a directed acyclic graph, and the node set includes observable variables, hidden variables and the construction entropy as a macro state variable.

[0011] Preferably, the causal state evolution model describes its dynamic evolution mechanism through a state transition function, which defines a specific conditional dependence function for each non-root node in the graph. The conditional dependence function includes at least one of an analytical model based on physical mechanism, a regression model based on data-driven or a conditional probability distribution based on probability theory.

[0012] Preferably, the steps of predicting and evaluating the construction progress disturbance risk include: Define a multi-dimensional composite discrimination criterion including the amplitude, growth rate and duration of the entropy; The construction entropy evolution trend is compared with the multi-dimensional composite discrimination criterion to identify risks. A probability mapping model is used to quantize the risk identification result into a progress disturbance probability value to determine the risk level.

[0013] Preferably, the step of generating and recommending the optimized rectification decision first generates a candidate rectification measure set matched with the current risk source from a preset rectification measure knowledge base.

[0014] Preferably, the step of generating and recommending the optimized rectification decision further comprises: The causal state evolution model is used to forward deduce each measure in the candidate rectification measure set to respectively evaluate the expected construction entropy evolution trend after the execution of the measure. An integrated utility function is used to quantitatively evaluate the expected rectification effect and rectification cost of each measure, and the candidate measure with the maximum integrated utility value is determined as the optimized rectification decision.

[0015] An intelligent cap beam jacking construction progress intelligent prediction system, comprising: A data acquisition and preprocessing module is configured to acquire multi-source real-time data in a cap beam jacking construction process and preprocess the multi-source real-time data to obtain a standardized state data set. A construction entropy calculation module is connected with the data acquisition and preprocessing module and is configured to calculate a construction entropy for measuring construction system uncertainty based on the standardized state data set. A model prediction module is connected with the construction entropy calculation module and is configured to establish a causal state evolution model and deduce and predict a construction entropy evolution trend in a future period based on the current construction system state and the causal state evolution model. A risk assessment module is connected with the model prediction module and is configured to predict and evaluate a construction progress disturbance risk in the future period based on the construction entropy evolution trend. A rectification decision module is connected with the risk assessment module and the model prediction module and is configured to generate and recommend an optimized rectification decision based on the causal state evolution model when the evaluation result of the construction progress disturbance risk exceeds a preset risk level.

[0016] The present application provides an intelligent cap beam jacking construction progress intelligent prediction method. The method has the following advantages: 1、The present application collects four kinds of multi-source real-time data of equipment, structure, personnel and environment, and constructs a construction entropy index capable of comprehensively quantifying system uncertainty based on information entropy theory. This method condenses complex and multi-dimensional construction state information into a single and dynamic macro quantitative index, overcomes the limitations of traditional methods relying on single parameter threshold or subjective experience judgment, and makes the overall risk assessment of the construction system more objective, comprehensive and accurate.

[0017] 2、The present application establishes a causal state evolution model including observable variables, hidden variables and macro state variables, and uses the model to deduce and predict the evolution trend of construction entropy in the future period. This prediction mechanism based on causal relationship not only reveals the evolution path of risk, but also identifies potential construction progress disturbance risks in advance, changes risk management from passive response to proactive early warning, and saves valuable decision-making and intervention time for managers.

[0018] 3、When the construction progress is predicted to have disturbance risks, the present application can perform counterfactual deduction based on the causal state evolution model or the sparse model after dynamic pruning, and generate an adaptive correction strategy that can actively reduce future construction entropy. This method replaces the emergency disposal mode relying on artificial experience through virtual simulation and effect evaluation of different correction measures, which helps to improve the scientific nature, pertinence and effectiveness of correction decision-making. BRIEF DESCRIPTION OF DRAWINGS

[0019] Figure 1 is a method flowchart of the present application; Figure 2 is a construction entropy calculation process diagram of the present application; Figure 3 is a causal state evolution model topological structure diagram of the present application; Figure 4 is a system architecture diagram of the present application. DETAILED DESCRIPTION

[0020] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the specification of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0021] Referring to the drawings Figure 1 - the drawings Figure 3 , Figure 1 is a cover beam intelligent lifting construction progress intelligent prediction method flowchart according to an embodiment of the present application. The present application provides a cover beam intelligent lifting construction progress intelligent prediction method, comprising the following steps: S10, collect multi-source real-time data in the process of the roof beam jacking construction, and pre-process the multi-source real-time data to obtain a standardized state data set; S20, based on the standardized state data set, construct and calculate construction entropy for measuring uncertainty of the construction system; S30, establish a causal state evolution model including observable variables, hidden variables and macro state variables, and based on the state of the construction system at the current time, use the model to deduce and predict the evolution trend of the construction entropy in the future period; S40, real-time monitor the value of the construction entropy and its change rate, when the value of the construction entropy or its change rate exceeds a preset threshold, dynamically prune the causal state evolution model to generate a sparse model focusing on the core causal path; S50, when the evolution trend of the construction entropy indicates that there is a disturbance risk in the construction progress, perform counterfactual reasoning based on the causal state evolution model or the sparse model to generate an adaptive correction strategy that can reduce the construction entropy in the future period.

[0022] In order to make the purpose, technical scheme and advantages of the present application clearer, the technical scheme of the present application will be described in detail below.

[0023] S10 step, that is, collecting multi-source real-time data in the process of the roof beam jacking construction, and pre-processing the multi-source real-time data to obtain a standardized state data set, specifically including the following sub-steps: S101, through the data acquisition system deployed at the construction site, real-time acquire four types of original data reflecting the state of the roof beam jacking construction. The multi-source real-time data explicitly include but are not limited to: Device state data: collected through sensors installed on the hydraulic pump station and each synchronous jacking jack. For example, use a pressure transmitter to acquire the real-time pressure of the hydraulic main oil circuit and each branch oil circuit; use a linear variable differential transformer (LVDT) or a wireline displacement sensor to measure the accurate extension amount of each jack piston; and use a thermocouple or a thermistor sensor to monitor the temperature of the hydraulic oil and key components of the device.

[0024] Structure state data: collected through health monitoring sensors pre-deployed at key positions of the roof beam body. For example, use a resistance strain gauge to measure the stress change of the key stress point of the roof beam in the jacking process; use a static water level gauge or a laser range finder to monitor the attitude of the roof beam in space, including its absolute elevation, plane displacement and inclination.

[0025] Personnel operation data: Obtained by recording the operation log of the human-machine interface (HMI) or programmable logic controller (PLC) of the central control console. This data specifically includes the timestamp of the operation instruction, the type of instruction (such as start, pause, fine-tune, stop), the instruction parameters (such as target pressure value, target displacement value), and the operator's revocation or modification of instruction records.

[0026] Environmental state data: Collected through a small weather station or dedicated environmental sensors. For example, use an anemometer to measure the real-time wind speed and direction at the site; use a thermometer and hygrometer to measure the environmental temperature and air humidity, respectively.

[0027] All the above raw data are continuously collected at a preset sampling frequency (e.g. 1 Hz) and transmitted to the central data server for storage through industrial wireless network or wired bus technology.

[0028] S102, pre-processing operations are performed on the raw data stored in the central data server to generate standardized state data sets that can be directly used for subsequent entropy calculation and model input. The pre-processing operations include data cleaning, time series alignment and data normalization in turn.

[0029] Data cleaning aims to identify and process abnormal data in the raw data sequence. Specifically, the 3σ criterion can be used to judge the effectiveness of the data points, and the data points that exceed the range of the mean plus or minus three times the standard deviation are identified as outliers.

[0030] For the identified outliers or missing values caused by signal transmission interruption, linear interpolation or spline interpolation method can be used for filling to ensure the integrity of the data sequence.

[0031] Time series alignment aims to unify data streams from different sensors, with different sampling frequencies or transmission delays, to the same time reference. Specifically, a standard time interval (e.g. 1 second) is set, and all data sequences are aligned to the time points of the standard interval through resampling technology. For data with a sampling frequency higher than the standard interval, the mean or the last value within an interval can be taken as a representative; for data with a sampling frequency lower than the standard interval, the value of the previous valid data point can be used for filling.

[0032] Data normalization aims to eliminate the numerical differences caused by different physical units and dimensions of each state parameter, and map all data to a unified interval. Specifically, the min-max normalization method can be used to convert the original value of any parameter to a normalized value. The conversion formula is: ; In the formula, is the normalized parameter value; is the real-time measurement value of the parameter. is the minimum value of the parameter within the historical data or the preset safety range; is the maximum value of the parameter within the historical data or the preset safety range.

[0033] After normalization, the values of all parameters are mapped to the interval [0, 1] to form a final standardized state data set for subsequent steps S20 and S30.

[0034] In this embodiment, by systematically collecting four types of data of equipment, structure, personnel and environment on the construction site, and sequentially performing data cleaning, time alignment and normalization processing, the transformation of multi-source heterogeneous original monitoring data into standardized state data set is realized, which helps to lay a solid data foundation for the accurate quantification of subsequent construction entropy and reliable input of causal state evolution model.

[0035] S20, that is, based on the standardized state data set, a construction entropy for measuring the uncertainty of the construction system is constructed and calculated, which specifically includes the following sub-steps: S201, defining the composition of construction entropy. Construction entropy is a quantitative representation of the overall disorder or uncertainty of the construction system at a certain time. Its value is obtained by weighted summation of four independent sub-state entropies, i.e. equipment state entropy, structure state entropy, personnel state entropy and environment state entropy, through a preset weight coefficient. The relationship is determined by the following formula: ; In the formula, is the construction entropy; is the weight coefficient of equipment state entropy; is the equipment state entropy; is the weight coefficient of structure state entropy; is the structure state entropy; is the weight coefficient of personnel state entropy; is the personnel state entropy; is the weight coefficient of environment state entropy; is the environment state entropy.

[0036] The determination of weight coefficients , , and is not arbitrary, but is determined by one or more systematic methods according to the risk characteristics of the specific project to ensure its objectivity and reasonableness.

[0037] One specific implementation is to use the analytic hierarchy process (AHP), and organize field experts to compare and score the contribution of four factors, i.e., equipment, structure, personnel and environment, to the overall construction risk, build a judgment matrix, and obtain the weight of each factor by calculating the maximum eigenvalue and eigenvector of the matrix.

[0038] Another implementation is to quantify the contribution of each factor to the historical problems through sensitivity analysis or regression analysis based on the failure or delay data of the same type of historical projects, and allocate the weight based on this.

[0039] S202, calculate each sub-state entropy. The calculation process of each sub-state entropy follows a unified methodology, which includes calculating parameter deviation in a sliding time window, discretizing the deviation value, calculating state probability, and applying the information entropy formula. Taking the calculation of equipment state entropy as an example, the specific implementation is as follows: First, set a sliding time window (e.g., the last 5 minutes), and for each equipment parameter extracted from the standardized state data set within the window, calculate its deviation sequence relative to the baseline value. The baseline value is determined according to the nature of the parameter, for example, for the pressure of a jack, the baseline value is the target pressure set value under the current working condition; for the temperature of hydraulic oil, the baseline value is the midpoint of the ideal temperature interval for normal operation.

[0040] Second, discretize the calculated deviation value sequence, i.e., map the continuous deviation value to a pre-set discrete state interval. The specific implementation methods of discretization include but are not limited to: equal-width discretization, i.e., divide the entire possible range of deviation values (e.g., the normalized [0,1] interval) into intervals with equal width; equal-frequency discretization, i.e., adjust the width of each interval so that the number of data points falling into each interval is roughly equal; discretization based on clustering algorithm, for example, using K-Means clustering algorithm to automatically cluster the deviation value data into clusters, and the range of each cluster defines a discrete state interval.

[0041] Third, calculate the frequency of each item of equipment parameter falling into the above K discrete state intervals within the time window, and calculate the corresponding state probability according to the frequency. For the th equipment parameter, the probability of its deviation value falling into the th state interval is determined by the frequency of the state divided by the total number of data points in the time window.

[0042] Finally, by applying the information entropy formula, the information entropy of all device parameters under all discrete states is summed to obtain the total value of the device state entropy, as shown in the following formula: ; In the formula, The device state entropy; An index for device parameters; The total number of equipment monitoring parameters included in the calculation; This is the index of the discrete state interval; The total number of discrete state intervals; For the first The deviation value of the device parameter falls within the first... The probability of each state interval.

[0043] To further clarify, the calculation process for other sub-state entropies also follows this methodology. For example, when calculating the structural state entropy, the baseline value for its parameter deviation is derived from the theoretical stress or displacement calculation values ​​of the finite element analysis (FEM) model for the current working condition.

[0044] When calculating the entropy of a person's state, the deviation of its parameters (such as operation response time) is obtained by comparing it with the average response time of the operator executing similar instructions under normal, non-emergency conditions. In this way, the universality and effectiveness of entropy calculation across various data sources are ensured.

[0045] In step S203, the four sub-state entropies calculated in S202 are substituted into the weighted summation formula defined in S201 to obtain the construction entropy at the current moment, which comprehensively reflects the macroscopic uncertainty level of the entire construction system. This construction entropy value will be used as a key dynamic indicator and input into subsequent steps S30 and S40.

[0046] This implementation method achieves a single, dynamic quantitative representation of the overall uncertainty of a complex construction system by constructing multi-dimensional sub-state entropies and weighting and fusing them.

[0047] Step S30 involves establishing a causal state evolution model that includes observable variables, latent variables, and macroscopic state variables. Based on the current state of the construction system, the model is used to deduce and predict the evolution trend of construction entropy in future periods. This step specifically includes the following sub-steps: S301, constructing the static topology of the causal state evolution model. Mathematically, the causal state evolution model is defined as a directed acyclic graph, whose node set contains all the key state variables characterizing the jacking construction process of the cap beam. These variables are divided into three categories: Observable variables are physical quantities that can be directly obtained through sensors, such as the pressure of a specific jack or the stress at a certain measuring point on a cap beam. Hidden variables, i.e. latent factors that cannot be directly measured but have a key impact on the system state evolution, such as the cumulative wear degree of a specific equipment or the fatigue index of an operator, whose values can be estimated by a weighted combination of multiple relevant observable variables or a mechanism-based empirical model; Macro state variables, i.e. the construction entropy and its four sub-state entropies calculated in the preceding steps, which are used to characterize the overall uncertainty of the system. The edge set of the graph represents the objectively existing direct causal relationship between the node variables. A directed edge from one node to another indicates that the former is the direct cause of the latter.

[0048] The topological structure of the causal graph can be achieved by combining the following two ways: Firstly, based on the knowledge of domain experts, the physical laws, process flow and management experience of the bent cap jacking construction are transformed into the initial causal network structure. Secondly, causal discovery algorithms such as PC algorithm or LiNGAM algorithm are applied to learn and verify the conditional independence relationship between variables from a large amount of state data accumulated in historical construction projects, so as to modify and confirm the graph structure constructed by expert knowledge.

[0049] S302, define the dynamic evolution mechanism of the causal state evolution model. The dynamics of the model is described by a state transition function, which defines the evolution rule of the system state vector from the current time to the next time.

[0050] ; In the formula, is the state vector of the system at the next time; is the state transition function; is the state vector containing the numerical values of all node variables in the graph at the current time; is the vector of external input or human intervention, such as a management instruction or a sudden environmental change; is the parameter set of the model.

[0051] The parameter set specifically defines the conditional dependence relationship of each node in the graph. For any node in the graph, its value at the next time is calculated by the values of all its parent nodes at the current time through a specific function. The specific form of the function depends on the nature of the causal relationship it describes, which can be implemented in one or more of the following ways: Analytical model based on physical mechanism: for the causal relationship between variables with clear physical laws, the function is an analytical expression. For example, in calculating the stress of a certain measuring point of the bent cap, the corresponding function can be a response surface model or a simplified mechanics equation based on structural mechanics theory, and the input of the equation is its parent node, i.e. the real-time pressure of each jack.

[0052] The above model can be simulated and calibrated in advance by finite element analysis software to obtain accurate function form and parameters.

[0053] Data-driven regression model: for causal relationships with complex physical mechanisms or difficult to accurately model, the function is a statistical regression model learned from historical data. For example, in estimating the hidden variable "operator fatigue index", the corresponding function can be a multivariate nonlinear regression model, such as a multilayer perceptron neural network with continuous working time of the operator, instruction modification frequency, etc. as input. The weights and biases of this neural network, i.e. the model parameters, are obtained by training on historical operation log data.

[0054] Conditional probability distribution based on probability theory: when the model is constructed as a Bayesian network, the function is a conditional probability distribution. For example, the conditional probability distribution of the node "hydraulic system state" (which can take discrete states such as "normal", "mild abnormality", "serious abnormality", etc.) defines the probability of the node being in each of its discrete states given that its parent nodes "hydraulic oil temperature" and "main pump outlet pressure" are in different state intervals. In this case, the value of the node at the next time step will be determined by random sampling according to this conditional probability distribution.

[0055] S303, forward reasoning based on the causal state evolution model to predict the construction entropy evolution trend in the future period, the prediction process takes the system state vector obtained after processing in the previous steps as the initial condition.

[0056] Subsequently, by iteratively applying the state transition function defined in S302, the predicted state vector sequence at multiple future time steps is calculated in turn. Since the construction entropy is a component of the state vector, the construction entropy prediction value sequence at multiple future time steps is also obtained in the prediction sequence. This sequence is a quantitative prediction of the evolution trend of the construction system uncertainty in the future period, which will be passed to the subsequent steps as the basis for risk judgment and correction decision.

[0057] This embodiment realizes the transformation of discrete, multi-dimensional construction state data into a system evolution model that can be dynamically and forwardly deduced by constructing a causal relationship network that integrates expert knowledge and data-driven methods and defining a rigorous state transition function. This helps to provide a solid model foundation for subsequent construction progress disturbance risk prediction and intelligent correction decision-making.

[0058] S40 step, i.e. predicting and evaluating the construction progress disturbance risk in the future period based on the construction entropy evolution trend, which includes the following sub-steps: S401, define the composite discriminant criterion of construction progress disturbance risk. The criterion is not based on a single threshold, but is comprehensively judged by a multi-dimensional condition set containing the amplitude, growth rate and duration of entropy.

[0059] Specifically, when the predicted future construction entropy evolution trend in S30 step meets any of the following conditions, it is determined that there is a construction progress disturbance risk in the future: Amplitude over-limit criterion: the value of the predicted construction entropy at a certain time in the future exceeds the maximum safety entropy threshold representing the absolute unacceptable state of the system.

[0060] Growth rate over-limit criterion: the growth rate of the predicted construction entropy within a certain period in the future exceeds the critical growth rate threshold representing the rapid deterioration of system stability.

[0061] Continuous anomaly criterion: the value of the predicted construction entropy is continuously higher than a warning entropy threshold representing the non-ideal state of the system for a continuous period of time, and the duration exceeds the maximum tolerance duration allowed.

[0062] The maximum safety entropy threshold, critical growth rate threshold, warning entropy threshold and maximum tolerance duration are pre-set according to the safety data statistics of similar historical projects, simulation experiment results and experience of experts in the field, and are stored in the system configuration library.

[0063] S402, based on the predicted construction entropy sequence, perform forward-looking risk identification, first, the system obtains the predicted construction entropy value sequence of multiple time steps in the future from S30 step.

[0064] Then, the system analyzes each predicted value and its change trend in the sequence one by one, and compares it with the composite discriminant criterion defined in S401 in real time. Once it is found that any one of the time points or time periods in the sequence meets any of the above amplitude over-limit, growth rate over-limit or continuous anomaly criteria, the system generates a construction progress disturbance risk warning signal. The signal not only identifies the existence of risk, but also records the specific criterion type that triggers the warning and the future time point of the predicted risk.

[0065] S403, quantitative evaluation of the identified risk to determine its severity level, in order to avoid the information loss caused by simple binary (risk / no risk) judgment, further introduce the calculation of progress disturbance probability to quantify the risk, the calculation of the probability can be realized through a pre-trained probability mapping model, the formula is as follows: ; In the formula, is the predicted construction progress disturbance probability, its value range is [0, 1]; The index of the criterion, respectively corresponding to the amplitude, growth rate and duration of the criterion; The weight coefficient corresponding to the first The weight coefficient corresponding to the first The normalized overrun of the first The normalized overrun of the first criterion, for example, for the amplitude criterion, the overrun is the degree to which the predicted entropy value exceeds the maximum safe entropy threshold; A calibration constant for adjusting the decision boundary of the function.

[0066] The progress disturbance probability value obtained by calculation can be used to classify the risk, for example, the probability value above 0.75 is defined as "high risk" and needs immediate intervention; the probability value between 0.5 and 0.75 is defined as "medium risk" and needs attention; the probability value below 0.5 is defined as "low risk" and only needs routine monitoring.

[0067] In this embodiment, by setting the multi-dimensional composite discriminant criterion of the amplitude, growth rate and duration of the entropy, and using the predicted construction entropy evolution trend for forward-looking assessment, the automatic identification and quantitative assessment of the future construction progress disturbance risk are realized, which helps to change the risk management from post-response to pre-warning, and provides accurate trigger time and risk level basis for subsequent correction decision-making.

[0068] S50 step, that is, when the construction progress disturbance risk is predicted, the optimized correction decision is generated and recommended, which specifically includes the following sub-steps: S501, based on the pre-set correction measure knowledge base, a candidate correction measure set is generated. The correction measure knowledge base is a structured database that pre-stores standardized operation procedures verified for different risk sources. The construction of the knowledge base is based on construction design specifications, emergency plans and historical experience of field experts.

[0069] When the S40 step outputs a specific risk warning signal (for example, "high risk" warning caused by "specific jack pressure anomaly"), the system will automatically retrieve and match all correction measures related to the risk source from the knowledge base to form a set containing one or more candidate measures. For example, for pressure anomaly, the candidate measures can include: "Reduce the loading rate of all jacks by 20%", "Suspend construction and check the pressure sensor", "Perform micro-unloading operation on the jack with pressure anomaly", etc.

[0070] S502, using the causal state evolution model, forward inference is performed on each measure in the candidate set of corrective measures to evaluate its expected corrective effect. For each candidate measure, the system inputs it as a new external intervention vector into the causal state evolution model established in S30.

[0071] Subsequently, the model will reiterate the calculation with the current system state as the initial condition, generating a new construction entropy evolution prediction trajectory in the future period after the implementation of the measure. This process is equivalent to a "virtual execution" or "digital twin" simulation of each candidate measure.

[0072] S503, the expected corrective effect of each candidate measure is quantitatively evaluated by a comprehensive utility function, and the optimal corrective decision is determined. The comprehensive utility function aims to score each predicted entropy evolution trajectory from the pros and cons of the corrective effect and the high and low of the corrective cost. A specific utility function can be defined as: ; In the formula, is the comprehensive utility value of the candidate measure; and are the weight coefficients representing the corrective effect and the corrective cost, respectively; is the predicted time range; is a reference entropy value representing the ideal stable state (usually a small value close to zero); is the predicted construction entropy value at future time after the implementation of the measure ; is the cost brought by the implementation of the measure .

[0073] The system will calculate the corresponding comprehensive utility value for each candidate measure. The higher the utility value, the faster and more effectively the measure can theoretically suppress the future uncertainty of the system (construction entropy) at a lower cost to the ideal level.

[0074] S504, the candidate measure with the highest comprehensive utility value is output and recommended as the optimal corrective decision. The system will package the measure with the highest calculated utility value into a clear and executable instruction and present it to the operators in the field control room through the human-machine interaction interface (HMI).

[0075] ​The recommended information not only contains specific operation content (for example, "it is recommended to reduce the loading rate of the No. 2 and No. 3 jacks by 15%"), but also carries its expected correction effect (for example, "it is expected to reduce the construction entropy below the early warning threshold within 5 minutes") and the corresponding risk assessment (for example, "the operation has little effect on the total construction period"), thereby providing comprehensive and transparent decision support for the operator.

[0076] In this embodiment, by constructing a correction measure knowledge base and using a causal state evolution model to perform forward deduction and utility evaluation on the candidate measures, intelligent and optimized correction decision recommendation is realized from risk early warning, which helps to combine human experience with accurate model calculation, improve the scientificity and effectiveness of emergency response, and finally form a closed-loop control of construction risk management.

[0077] Referring to the accompanying drawings Figure 4 The present application provides a cap beam intelligent jacking construction progress intelligent prediction system, comprising: A data acquisition and preprocessing module is used to acquire multi-source real-time data in the cap beam jacking construction process and preprocess the multi-source real-time data to obtain a standardized state data set. A construction entropy calculation module is connected with the data acquisition and preprocessing module and is used to calculate the construction entropy for measuring the uncertainty of the construction system based on the standardized state data set. A model prediction module is connected with the construction entropy calculation module and is used to establish a causal state evolution model and deduce and predict the construction entropy evolution trend in the future period based on the current construction system state by using the causal state evolution model. A risk assessment module is connected with the model prediction module and is used to predict and evaluate the construction progress disturbance risk in the future period based on the construction entropy evolution trend. A correction decision module is connected with the risk assessment module and the model prediction module and is used to generate and recommend the optimized correction decision based on the causal state evolution model when the evaluation result of the construction progress disturbance risk exceeds the preset risk level.

Claims

1. A method for intelligent prediction of the construction progress of intelligent jacking of cap beams, characterized in that, Includes the following steps: Collect multi-source real-time data during the construction process of lifting the cap beam, and preprocess the multi-source real-time data to obtain a standardized state dataset; Based on the standardized state dataset, a construction entropy is constructed and calculated to measure the uncertainty of the construction system. A causal state evolution model is established, which includes observable variables, latent variables, and macroscopic state variables. Based on the current state of the construction system, the model is used to deduce and predict the evolution trend of construction entropy in future periods. The value of the construction entropy and its rate of change are monitored in real time. When the value of the construction entropy or its rate of change exceeds a preset threshold, the causal state evolution model is dynamically pruned to generate a sparse model focused on the core causal path. When the evolution trend of the construction entropy indicates that there is a risk of disturbance to the construction progress, counterfactual inference is performed based on the causal state evolution model or the sparse model to generate an adaptive correction strategy that can reduce the construction entropy in future periods.

2. The intelligent prediction method for the construction progress of intelligent jacking of cap beams according to claim 1, characterized in that, The multi-source real-time data includes: Equipment status data, structural status data, personnel operation data, and environmental status data.

3. The intelligent prediction method for the construction progress of intelligent jacking of cap beams according to claim 1, characterized in that, The construction entropy is obtained by weighting and summing the equipment state entropy, structural state entropy, personnel state entropy, and environmental state entropy using preset weighting coefficients.

4. The intelligent prediction method for the construction progress of intelligent jacking of cap beams according to claim 3, characterized in that, The calculation method for any one of the equipment state entropy, structural state entropy, personnel state entropy, and environmental state entropy includes: Within the sliding time window, for the multiple parameters corresponding to the state entropy, calculate the deviation sequence of each parameter relative to its reference value; The deviation sequence is discretized to obtain multiple discrete state intervals; Statistically calculate the state probabilities of the deviation sequence within the multiple discrete state intervals; The state entropy value is calculated by applying the information entropy formula based on the state probability.

5. The intelligent prediction method for the construction progress of intelligent jacking of cap beams according to claim 1, characterized in that, The causal state evolution model is a directed acyclic graph, and the node set includes observable variables, latent variables, and the construction entropy as a macroscopic state variable.

6. The intelligent prediction method for the construction progress of intelligent jacking of cap beams according to claim 5, characterized in that, The causal state evolution model describes its dynamic evolution mechanism through a state transition function. The state transition function defines a specific conditional dependency function for each non-root node in the graph. The conditional dependency function includes at least one of the following: an analytical model based on physical mechanisms, a data-driven regression model, or a conditional probability distribution based on probability theory.

7. The intelligent prediction method for the construction progress of intelligent jacking of cap beams according to claim 1, characterized in that, The steps for predicting and assessing the risk of construction schedule disruption include: Define a multi-dimensional composite discrimination criterion that includes the magnitude, growth rate, and duration of entropy; The evolution trend of construction entropy is compared with the multi-dimensional composite discrimination criterion to identify risks; The risk identification results are quantified into schedule disturbance probability values ​​through a probability mapping model to determine the risk level.

8. The intelligent prediction method for the construction progress of intelligent jacking of cap beams according to claim 1, characterized in that, The step of generating and recommending the optimal corrective decision first involves generating a set of candidate corrective measures that match the current source of risk from a pre-set knowledge base of corrective measures.

9. The intelligent prediction method for the construction progress of intelligent jacking of cap beams according to claim 8, characterized in that, The step of generating and recommending the optimal corrective decision also includes: Using the causal state evolution model, a forward extrapolation is performed on each measure in the candidate corrective measures set to evaluate the expected construction entropy evolution trend after its implementation. Furthermore, a comprehensive utility function is used to quantitatively evaluate the expected corrective effects and costs of each measure, and the candidate measure with the highest comprehensive utility value is determined as the optimal corrective decision.

10. An intelligent prediction system for the construction progress of a girder intelligent jacking project, comprising the intelligent prediction method for the construction progress of a girder intelligent jacking project according to any one of claims 1-9, characterized in that, include: The data acquisition and preprocessing module is used to acquire multi-source real-time data during the construction process of the cap beam lifting, and to preprocess the multi-source real-time data to obtain a standardized state dataset. The construction entropy calculation module, connected to the data acquisition and preprocessing module, is used to calculate the construction entropy, which measures the uncertainty of the construction system, based on the standardized state dataset. The model prediction module, connected to the construction entropy calculation module, is used to establish a causal state evolution model and, based on the current state of the construction system, to deduce and predict the construction entropy evolution trend in future periods using the causal state evolution model. The risk assessment module, connected to the model prediction module, is used to predict and assess the risk of construction progress disturbance in the future period based on the construction entropy evolution trend. The deviation correction decision module, connected to the risk assessment module and the model prediction module, is used to generate and recommend the optimal deviation correction decision based on the causal state evolution model when the assessment result of the construction progress disturbance risk exceeds the preset risk level.

Images