Engineering project full life cycle dynamic scheduling method and system

Abstract

The invention relates to the technical field of engineering project management, and discloses an engineering project full life cycle dynamic scheduling method and system, and the method comprises the steps: analyzing BIM model data, discretizing geometric components into BIM space-time volume elements, endowing the BIM space-time volume elements with potential energy values, and constructing a three-dimensional construction potential energy field; constructing a multi-dimensional hybrid constraint boundary in combination with gravity support, operation space and resource stock; generating a multi-fork evolutionary decision tree by taking the current field state as a root node; defining a scheduling function which takes maximization of the space-time entropy reduction rate of the system as a target, and performing efficiency evaluation and dead branch fusing on branch paths by taking the scheduling function as a pruning operator; and extracting a dynamic scheduling priority index of the to-be-executed voxel in the optimal path, and generating an instruction to drive on-site resource configuration and operation. By constructing the construction potential energy field and the entropy subtraction decision tree, adaptive dynamic scheduling under resource fluctuation and space conflicts is realized, and global optimization of space-time efficiency of an engineering system is ensured.

Description

Technical Field

[0001] This invention relates to the field of engineering project management technology, and more specifically, to a method and system for dynamic scheduling of the entire life cycle of an engineering project. Background Technology

[0002] As modern construction engineering evolves towards super high-rise buildings, deep foundation pits, and large-scale complexes, the full lifecycle management of engineering projects has become a complex systems engineering project involving massive data interaction and multi-dimensional resource collaboration. In the current digital construction system, Building Information Modeling (BIM) technology, by constructing a three-dimensional digital foundation, realizes the digital expression of building geometry and physical attributes, providing a high-precision static spatial reference for the construction process. In actual project management, the scheduling system acts as the "central nervous system," needing to coordinate the overlap of work processes in the time dimension, the allocation of work surfaces in the spatial dimension, and the supply of human, machine, and material resources in the physical dimension. The entire project lifecycle covers a long chain from detailed design, material procurement, on-site construction to final delivery, with highly non-linear coupling relationships between each link. Any slight schedule deviation or resource fluctuation can have a butterfly effect in subsequent stages. Therefore, how to achieve real-time perception and precise allocation of construction tasks, on-site space, and supply chain resources in a dynamically changing physical environment is the core requirement for building a new generation of intelligent construction systems. Currently, on the one hand, traditional engineering scheduling methods mostly rely on Gantt charts or the Critical Path Method (CPM) for static planning. This linear extrapolation model assumes that the construction environment is ideal and constant, making it difficult to cope with complex spatiotemporal conflicts on site. Once an emergency occurs, it often has to rely on manual experience for local repairs, resulting in a serious disconnect between the plan and reality. On the other hand, although existing BIM 4D technology has achieved visualized simulation of progress, it is essentially still an animated demonstration with a preset timeline, lacking adaptive calculation capabilities based on the physical state of the site. It cannot automatically calculate the optimal solution when work surfaces overlap or resources are contested. In addition, existing resource management systems usually treat material supply as a deterministic event, ignoring the random interference brought about by supply chain fluctuations. They lack mathematical modeling and flexible fault tolerance mechanisms for uncertain factors, which makes the entire scheduling system prone to stagnation or disorder when resources are delayed in delivery. It cannot fundamentally solve the problem of inefficiency caused by the "spatiotemporal entropy increase" on the construction site. Summary of the Invention

[0003] The purpose of this invention is to provide a dynamic scheduling method and system for the entire life cycle of engineering projects. By constructing a construction potential energy field and an entropy reduction decision tree, and combining Bayesian resource correction and Nash equilibrium game, it achieves adaptive dynamic scheduling under resource fluctuations and spatial conflicts, ensuring the global optimal spatiotemporal efficiency of the engineering system.

[0004] This invention is achieved through the following technical solution: A dynamic scheduling method for the entire lifecycle of an engineering project, comprising the following steps: Access the project management database, load and read the project BIM model data, discretize the geometric components into BIM spatiotemporal elements, assign a construction potential energy value to each BIM spatiotemporal element, and construct a construction potential energy field in three-dimensional space. Identify the gravity support constraints and work surface space constraints that the BIM spatiotemporal elements are subject to during construction, and construct a multi-dimensional hybrid constraint boundary by combining the inventory of human, machine and material resources in the project management database. Using the current site status as the root node, within the multidimensional hybrid constraint boundary, the construction sequence of the BIM spatiotemporal elements is subjected to discrete state fission to generate a multi-branch evolutionary decision tree. A scheduling function is defined with the optimization objective of maximizing the spatiotemporal entropy reduction rate of the engineering system. This function is used as the pruning operator of the multi-branch evolutionary decision tree. The cumulative efficiency of each branch path is calculated in the future time window, and dead branches are broken for branches with cumulative efficiency below a set threshold. From the set of retained optimal paths, extract the dynamic scheduling priority index of each BIM spatiotemporal element to be executed at the current moment, generate instructions to drive the on-site construction resource allocation, and generate scheduling instructions to drive on-site operations.

[0005] Optionally, the construction of the construction potential energy field in three-dimensional space specifically includes: Extract the volume data and material properties of each geometric component from the engineering BIM model data, and calculate its engineering quantity weight; Based on the critical path method network graph, the spatiotemporal topological distance of each geometric component from the project delivery node is calculated, and the reciprocal of the spatiotemporal topological distance is defined as the urgency coefficient. Based on the first set formula, each BIM spatiotemporal element is assigned a construction potential energy value to construct a construction potential energy field in three-dimensional space.

[0006] Optionally, the first set formula is specifically calculated as follows:

[0007] in, Let be the construction potential energy value released by the i-th BIM spatiotemporal element. As the weight of the project quantity, For spatiotemporal topological distance, and This is the preset adjustment weighting factor.

[0008] Optionally, the construction of the multidimensional hybrid constraint boundary includes resource uncertainty correction based on Bayesian inference, specifically as follows: Retrieve resource supply logs from historical projects in the project management database, analyze the time deviation distribution of various resources from order placement to arrival, and establish a prior probability model. Furthermore, by using IoT gates and positioning tags deployed on-site, the actual entry records of the current batch of resources are collected as observational evidence. ; Based on Bayes' theorem Calculate the posterior probability distribution to obtain the confidence level of resource availability within a specific time window; Prior probability model With observational evidence The posterior probability distribution of resource supply delay is merged and updated, and the cumulative probability value of the posterior probability distribution within the preset safety interval is calculated as the availability confidence. An inverse proportional mapping function is established to transform the availability confidence into an elastic relaxation factor with a value in the range (0,1), so that the multi-branch evolutionary decision tree can shrink the branch search range during periods of risk in resource supply.

[0009] Optionally, the generation of the multi-branch evolutionary decision tree is specifically configured with a Nash equilibrium game mechanism, which is as follows: When performing discrete state fission, identify multiple competing BIM spatiotemporal elements vying for the same scarce resource under the current node; Construct a multi-agent non-cooperative game matrix, using the construction potential energy value of each BIM spatiotemporal element. The revenue parameter is used, while the delay penalty caused by waiting is used as the cost parameter. To find the Nash equilibrium solution of the game matrix, only the BIM spatiotemporal volume element sorting combination that is in equilibrium is used as an effective branch to generate child nodes, and the others are discarded.

[0010] Optionally, the defined scheduling function is optimized by maximizing the spatiotemporal entropy reduction rate of the engineering system, and its specific calculation formula is as follows:

[0011] in, For a branch path in a decision tree The cumulative spatiotemporal entropy reduction efficiency; k is the time step index, K is the total number of steps in the prediction time window; The set of BIM spatiotemporal volume elements executed in parallel within the k-th time step; The construction potential energy value released by the i-th BIM spatiotemporal element; This is the time sensitivity coefficient; For execution time; The current moment; Let be the resource consumption cost of the i-th individual element; This is the congestion penalty gain coefficient; Let be the set of volume element pairs that have spatial conflicts within the k-th time step; The bounding box overlap volume of the conflicting element pairs; This is the standard safe working space constant.

[0012] Optionally, the The calculation steps are as follows: Read the geometric center coordinates and construction method type of the BIM spatiotemporal elements of the parallel operation; Based on the construction method type, assign the dynamic operation expansion radius to the BIM spatiotemporal elements to generate a three-dimensional dynamic bounding box. The intersection state of the three-dimensional dynamic bounding boxes of different BIM spatiotemporal elements is detected using the separation axis theorem. If they intersect, calculate the geometric volume of the intersecting portion based on Boolean intersection operations, and use it as... Substitute the values ​​into the scheduling function.

[0013] Optionally, the step of performing dead branch circuit breaking on branches with cumulative performance below a set threshold specifically involves: Calculate the cumulative performance achieved by the current branch path; Estimate the maximum potential energy release value of the remaining unconstructed BIM spatiotemporal elements under theoretically unconstrained conditions. ; like Less than the currently found global optimal solution preset ratio If this branch path is considered a dead branch, the calculation of its subsequent nodes is immediately terminated. .

[0014] Optionally, the extraction of the dynamic scheduling priority index of each BIM spatiotemporal element to be executed at the current time specifically includes: Calculate the frequency of BIM spatiotemporal elements to be executed in all valid paths that have not been broken. ; Calculate the delay gradient of BIM spatiotemporal elements ; The dynamic scheduling priority index is calculated using a second set formula, wherein the second set formula is:

[0015] in, These are the weighting coefficients. It is a BIM spatiotemporal volume element.

[0016] A dynamic scheduling system for the entire lifecycle of an engineering project, comprising: The data parsing and potential energy field construction module connects to the engineering project management database, loads and reads the engineering BIM model data, discretizes the geometric components into BIM spatiotemporal elements, assigns a construction potential energy value to each BIM spatiotemporal element, and constructs a construction potential energy field in three-dimensional space. The constraint boundary and game theory calculation module identifies the gravity support constraints and work surface space constraints that the BIM spatiotemporal elements are subjected to during construction, and constructs a multi-dimensional hybrid constraint boundary by combining the inventory of human, machine and material in the project management database. The entropy reduction decision tree evolution engine takes the current engineering site state as the root node and performs discrete state fission on the construction sequence of the BIM spatiotemporal elements within the multidimensional hybrid constraint boundary to generate a multi-branch evolutionary decision tree. It defines a scheduling function with the optimization objective of maximizing the spatiotemporal entropy reduction rate of the engineering system as the pruning operator of the multi-branch evolutionary decision tree, traverses and calculates the cumulative efficiency of each branch path in the future time window, and performs dead branch melting on branches with cumulative efficiency lower than a set threshold. The closed-loop feedback and command terminal extracts the dynamic scheduling priority index of each BIM spatiotemporal element to be executed at the current moment from the set of retained optimal paths, generates commands to drive the on-site construction resource allocation, and generates scheduling commands to drive on-site operations.

[0017] The technical solution of the present invention has at least the following advantages and beneficial effects: On the one hand, this invention creatively transforms abstract project progress into a visualized energy release process by mapping discretized BIM spatiotemporal elements into construction potential energy fields with physical properties, enabling the scheduling system to automatically find the optimal construction path based on the principle of minimum potential energy in physics. On the other hand, this invention introduces a resource uncertainty correction mechanism based on Bayesian inference, which can transform fuzzy supply chain fluctuations into elastic relaxation factors of constraint boundaries in real time, forcing the decision tree to automatically avoid highly dependent tasks during periods of high resource risk, achieving flexible risk avoidance at the system level. This not only effectively solves the spatial conflict problem when multiple trades are working together, but also transforms high-dimensional calculation results into intuitive on-site execution instructions through the DSPI index, thereby achieving a globally optimal balance between resource utilization and time cost throughout the entire project lifecycle while ensuring construction safety. Attached Figure Description

[0018] Figure 1 A flowchart illustrating the dynamic scheduling method for the entire lifecycle of engineering projects provided by this invention; Figure 2 A schematic diagram illustrating the principle of the dynamic scheduling system for the entire lifecycle of engineering projects provided by this invention. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The components of the embodiments of the present invention described and shown in the accompanying drawings can generally be arranged and designed in various different configurations.

[0020] like Figure 1 , Figure 2 As shown in the illustration, this invention provides a method and system for dynamic scheduling throughout the entire lifecycle of an engineering project, using a large-scale convention and exhibition center project as an example. This convention and exhibition center project is characterized by its complex structure, tight schedule, and frequent cross-disciplinary operations, making it difficult for traditional project management methods to cope with its dynamic nature and complexity. The method and system described in this embodiment can achieve intelligent and adaptive dynamic scheduling of this project.

[0021] In this embodiment, the dynamic scheduling system for the entire lifecycle of this project is deployed in the project's smart construction site command center. Its hardware architecture includes a high-performance central data server, multiple database servers, and an IoT sensing network covering the entire site. The central data server can be a tower server equipped with a multi-core CPU and a high-performance GPU, used to handle core scheduling function calculations and decision tree evolution tasks. The database servers store the project's BIM model data, supply chain data, historical project data, and real-time generated scheduling instructions. The IoT sensing network includes facial recognition gates deployed at site entrances and exits, GPS positioning modules for material transport vehicles, RFID tags on key components, and fixed laser scanners and UAV-borne panoramic cameras for on-site real-scene modeling. The system's software includes four core functional modules: a data parsing and potential field construction module, a constraint boundary and game theory calculation module, an entropy reduction decision tree evolution engine, and a closed-loop feedback and command terminal. The specific logical steps are as follows: Step 1: Connect to the project management database to perform data analysis and construct the construction potential energy field.

[0022] In this embodiment, step one aims to transform the static BIM model into a dynamic physical field containing the inherent driving force of construction. The system first securely connects to the project management database of the convention center project through a data interface. This database integrates the original BIM model files provided by the design institute, such as Autodesk Revit format RVT files or industry standard IFC files.

[0023] The data parsing and potential energy field construction module automatically performs the BIM model parsing task. It traverses every geometric component in the BIM model, such as a floor slab, a structural column, or a section of ventilation duct. For each component, the system discretizes its geometry into uniformly sized units, called BIM spatiotemporal elements. The purpose of discretization is to transform components of different shapes and from different disciplines into standardized computational objects, facilitating subsequent unified potential energy calculations and state fission. Each BIM spatiotemporal element not only contains the geometric information, material properties, and ID code of its component, but is also assigned a timestamp and status bits to characterize its temporal sequence and construction status during the scheduling process.

[0024] Next, the system enters the construction phase of the construction potential energy field. The concept of the construction potential energy field draws inspiration from the gravitational field in physics. Its core idea is that the more critical and urgent a component is, the higher its inherent construction potential energy, and the more likely it is to be prioritized for "release" in scheduling decisions. The construction potential energy value of each BIM spatiotemporal element is calculated using a comprehensive formula.

[0025]

[0026] in, Let be the construction potential energy value released by the i-th BIM spatiotemporal element. As the weight of the project quantity, For spatiotemporal topological distance, and This is a preset adjustment weighting factor. In this formula, the first term is the project quantity weight. The system reads the component volume data and material density attributes corresponding to the BIM spatiotemporal element i, calculates its physical mass or engineering cost, and performs normalization processing. For example, a massive core tube shear wall naturally has a higher engineering quantity weight than an ordinary infill wall block, reflecting its importance in the project. The second item is the spatiotemporal topological distance. The system first loads the project's Critical Path Method (CPM) network diagram. This network diagram defines the logical dependencies between all construction tasks; for example, the foundation pouring must be completed before the main structure construction can proceed. The system uses graph theory algorithms to calculate the longest path length in the network diagram from the final "project delivery" endpoint to the construction task corresponding to the current BIM spatiotemporal element i. This length is not a physical distance, but a topological distance that combines process dependencies and expected project duration. A component located at the beginning of the critical path has the longest spatiotemporal topological distance, indicating that its completion is fundamental to a large amount of subsequent work and its urgency is extremely high. Therefore, its reciprocal is taken. As a urgency coefficient, the longer the distance, the higher the urgency. and This is a preset adjustment weighting factor, and the sum of the two is 1. In the early stages of the project, this can be... Set it to a higher value, such as 0.7, to prioritize tasks on the critical path; later in the project, it can be adjusted higher. The value is used to prioritize completing the remaining finishing work with a large amount of remaining work. By assigning a construction potential energy value to each BIM spatiotemporal element, a quantified construction potential energy field covering the entire three-dimensional space of the building is constructed.

[0027] Step 2: Construct a multidimensional hybrid constraint boundary and perform Bayesian correction.

[0028] In this embodiment, step two involves defining a realistic feasible region for the subsequent decision tree search. Construction activities are not arbitrary but are subject to numerous rigid and flexible constraints. The system needs to identify these constraints and mathematically represent them.

[0029] In the specific implementation of this embodiment, the system identifies two main categories of rigid constraints. The first category is gravity support constraints, which means that a component can only begin construction after the load-bearing component below it has been completed. This is directly derived from physical laws, and the system automatically extracts it by analyzing the overlapping relationships of components in the BIM model. The second category is work surface space constraints, which means that the installation of certain large components requires a large temporary space around them, and no other work can be carried out simultaneously in this space.

[0030] The system identifies flexible constraints, primarily the stock constraints of physical resources such as personnel, machinery, and materials. The system connects to the project management database and reads information such as the current tonnage of various specifications of steel bars in the warehouse, the capacity of the concrete mixing plant, and the number of certified welders on site, forming a resource vector.

[0031] In practice, simply considering current inventory is insufficient because resource supply itself involves significant uncertainty. Therefore, this embodiment introduces a Bayesian correction mechanism for resource supply delays, transforming uncertainty into a calculable risk parameter. This process is divided into four sub-steps: Step A: Obtain Delayed Data Sources. The system's data parsing module will deeply mine the company's historical project database, retrieving historical supply records from various suppliers (such as steel mills and concrete companies) selected for the convention center project over the past few years. Statistical analysis will be performed on the deviations between the "planned delivery time" and the "actual delivery time" in these records to form a prior probability distribution model regarding "delay." This represents an initial judgment based on historical experience. Meanwhile, IoT devices deployed at the construction site begin to function. When a new batch of concrete mixer trucks arrives, their GPS location data is transmitted to the system in real time; when a batch of steel structure components passes through the RFID scanning device at the entrance, their arrival time, accurate to the second, is recorded. This real-time collected data constitutes observational evidence used to correct prior judgments. .

[0032] Step B involves calculating the posterior confidence score. A subroutine within the entropy-reducing decision tree evolution engine continuously runs the Bayesian inference algorithm. This algorithm utilizes the well-known Bayesian theorem. The system dynamically fuses prior probabilities with real-time evidence. Understandably, historical data shows a concrete supplier's average delay is 30 minutes, but if the first three trucks this morning arrived on time, the system will immediately calculate a new, more optimistic posterior probability distribution. Subsequently, the system calculates the cumulative probability value of this updated probability distribution function within a preset safety interval (e.g., delay less than 15 minutes). This cumulative probability is defined as the availability confidence level of the resource in the next time window. A 95% confidence level means the system is extremely confident that the resource will arrive on time.

[0033] Step C: Generate the elastic relaxation factor. After obtaining the probability value of availability confidence, the system transforms it into an elastic relaxation factor with a value between 0 and 1 using an inverse proportional mapping function. This function can be a simple linear mapping or a sigmoid function. Its core logic is that the higher the availability confidence, the better. The closer the value is to 1, the lower the confidence level becomes. When the confidence level drops to, for example, 60% due to a series of delayed events, The value will decrease significantly, for example, to 0.5.

[0034] Step D: Modify constraint boundaries and implement control. Elastic relaxation factor. The ultimate purpose is to dynamically scale the theoretical available upper limit of resources. When generating the decision tree, the system uses a formula for each resource. Calculate its dynamic resource boundaries. Wherein, This refers to the physical inventory in the warehouse. For example, there might be 200 tons of type A steel bars in the warehouse, but due to recent delays by the supplier, the availability confidence has decreased. The value became 0.6. Therefore, in the subsequent decision-making simulation, the system will pretend that only... Tons of steel bars are available. The direct consequence of this approach is that, in the next step of the decision tree evolution, construction plans that require more than 120 tons of type A steel bars will be judged as infeasible due to "insufficient resources" and automatically eliminated by the algorithm. Thus, the system cleverly forces the decision to shift to construction paths that rely on other, more reliable resources, completing a closed loop from risk perception to proactive avoidance.

[0035] Step 3: Using the current state as the root node, introduce Nash equilibrium game to generate a multi-branch evolutionary decision tree.

[0036] In this embodiment, after establishing clear constraint boundaries, the system begins to perform core future state deduction. It uses the current on-site engineering state (which components have been completed, and how much resource is in stock) as the unique root node of the decision tree.

[0037] Starting from the root node, the system performs discrete state fission, that is, it deduces which construction tasks might be executed next. Under the premise of satisfying all constraints, multiple BIM spatiotemporal elements may be in a "constructible" state. If these elements do not compete for any resources, the system treats them as a parallel combination. However, more commonly, multiple elements awaiting construction will compete for the same scarce resource; for example, two different areas of a convention center project may need to call up the only large crawler crane on the same day. In this case, this embodiment introduces a Nash equilibrium game mechanism to resolve such resource conflicts. The system treats these BIM spatiotemporal elements competing for the same resource as players in a game. Each player has two strategy choices: immediate construction or waiting. The system constructs a multi-agent non-cooperative game matrix. The payoff parameter in the matrix is ​​the construction potential energy value that each element can release if it "immediately constructs" and successfully acquires the resource. The cost parameter, on the other hand, is the penalty for project delays that will result if "wait" or "failure to compete" is selected. This penalty value is related to the urgency coefficient of the element.

[0038] The constraint boundary and game calculation module solves for the Nash equilibrium of this game matrix. A Nash equilibrium describes a stable state where no player can gain a better payoff by unilaterally changing their strategy. The solved equilibrium may be a pure strategy solution (e.g., A constructs, B waits) or a mixed strategy solution. The system only sorts and combines elements in Nash equilibrium states, using them as valid child nodes to grow new branches of the decision tree. Resource conflict combinations that lead to vicious competition and mutual destruction are completely eliminated during the decision tree generation stage because they do not conform to Nash equilibrium, greatly improving search efficiency. In this way, a multi-branch evolutionary decision tree is generated, with the current state as the root and possible future construction sequences as branches.

[0039] Step 4: Define the scheduling function and perform pruning and circuit breaking.

[0040] In this embodiment, the generated multi-branch evolutionary decision tree may contain an astronomical number of branch paths, necessitating an efficient evaluation and selection mechanism; this is the core task of this embodiment. The system defines a complex scheduling function with the optimization objective of maximizing the spatiotemporal entropy reduction rate of the engineering system, and uses this function as a pruning operator to eliminate inferior paths. Its mathematical expression is as follows:

[0041] in, For a branch path in a decision tree The cumulative spatiotemporal entropy reduction efficiency is denoted as , which is the cumulative spatiotemporal entropy reduction efficiency. A higher score indicates a better construction sequence; k is the time step index, and K is the total number of steps in the prediction time window; The set of BIM spatiotemporal volume elements executed in parallel within the k-th time step; The construction potential energy value released by the i-th BIM spatiotemporal element; This is the time sensitivity coefficient; For execution time; The current moment; Let be the resource consumption cost of the i-th individual element; This is the congestion penalty gain coefficient; Let be the set of volume element pairs that have spatial conflicts within the k-th time step; The bounding box overlap volume of the conflicting element pairs; This is the standard safe working space constant.

[0042] In this specific implementation, to evaluate the merits of each potential construction path, the system executes a composite calculation process based on efficiency-cost trade-offs and spatial conflict penalties. This calculation process does not rely on a single indicator but instead synthesizes multi-dimensional engineering factors into a unique quantitative score through the following step-by-step logic. The calculation process mainly includes four stages: time window slice traversal, basic efficiency ratio calculation, spatial congestion damping correction, and full-time cumulative aggregation.

[0043] Phase 1: The system first slices the entire construction branch path to be evaluated into several consecutive time windows according to a preset time step (e.g., every half shift or every hour). The system then independently evaluates the construction status within each time window, aiming to capture the dynamic changes in the construction process.

[0044] Phase Two: For each specific time window, the system performs an "input-output" analysis, which consists of revenue calculation in the numerator and cost calculation in the denominator. The system first identifies all BIM spatiotemporal elements planned for parallel execution within the given time window. For each element, the system reads its pre-defined construction potential energy value (representing importance and urgency). Based on this, the system introduces a time value decay logic. The system calculates the time difference between the planned completion time of the element and the current decision-making time. Based on this time difference, the system uses a natural exponential decay mechanism to discount the potential energy value: the earlier the planned completion time and the closer it is to the current time, the higher the retention rate of its potential energy value; conversely, the later the planned completion time is postponed, the more severely its potential energy value is decayed during calculation. The system then sums the time-discounted potential energy values ​​of all parallel elements within the time window as the total effective output of that time window.

[0045] The system calculates all physical resources required to support the construction of all the aforementioned parallel elements within this time window. This includes the monetized or standard unit values ​​of labor hours, machine shifts, and material consumption. The system sums up these resource consumption values ​​to determine the total resource input for this time window.

[0046] The system divides the total effective output calculated above by the total resource input to obtain a basic value. This value represents the ideal construction cost-effectiveness without considering spatial conflicts.

[0047] Phase 3: To prevent blindly pursuing high output from causing overcrowding or even safety accidents on-site, the system introduces a penalty mechanism that can "reset" the score to zero, namely, the calculation of the space congestion damping coefficient: The system iterates through all voxel pairs of parallel jobs within the time window, checking for physical overlap in their bounding spaces. If overlap exists, the system calculates the volume of the overlapping portion and compares it to a preset standard safe job space volume. The system uses logarithmic amplification logic to process this comparison. That is, when the overlap volume is small, the calculated conflict value increases slowly; however, as the overlap volume approaches or exceeds the safety threshold, the calculated conflict value rises sharply. The system summarizes the conflict values ​​of all conflicting voxel pairs and multiplies them by a weighting coefficient representing the severity of safety management.

[0048] Subsequently, the system uses negative exponential function logic to map this aggregated conflict value to a coefficient between 0 and 1. The correspondence is as follows: when there is no spatial conflict on site, the coefficient is 1 (i.e., no points are deducted); when the conflict is minor, the coefficient is slightly less than 1; when the conflict is severe, the coefficient approaches 0 very quickly. This coefficient is the spatial congestion damping.

[0049] The fourth stage is divided into single-step correction and full-time accumulation. Single-step correction involves multiplying the "basic efficiency-cost ratio" calculated in the second stage by the "space congestion damping" calculated in the third stage. This step implements a "one-vote veto" logic: no matter how large the construction output is in a given time period, if space congestion causes the damping coefficient to approach 0, the final score for that time period will be forcibly lowered to near zero, thus simulating the devastating impact of a safety accident on the schedule.

[0050] Accumulated over all time periods: The system sums up the final scores calculated for all time windows along the branch path. This sum is the final evaluation value (i.e., the total entropy reduction efficiency) for that construction path.

[0051] In this embodiment, The calculation steps are also carefully designed: the entropy reduction decision tree evolution engine reads the BIM data and construction methods of two voxels in parallel operations. For example, one is high-altitude welding, and the other is ground hoisting. Based on the construction method specifications, the system generates a dynamic 3D safety bounding box for each voxel. This bounding box is larger than the component itself, representing its actual operational impact range. Then, the system uses the Separating Axis Theorem (SAT) from computer graphics for fast collision detection. If the projections of two bounding boxes do not overlap on any coordinate axis, they are definitely not intersecting. If the projections of all axes overlap, they may intersect. Once a possible intersection is detected, the system further calls the geometric calculation program to perform a Boolean intersection operation on the two bounding boxes, accurately calculating the geometric volume of the overlapping part. This volume is... . This is a standard safe working space constant, which is understandable, 5 cubic meters. When there is no overlap, =0, The entire damping term is It doesn't work. Once overlap occurs, The term being greater than 0 causes the entire The value of this term is less than 1. Furthermore, the larger the overlap volume, the more exponentially the value of this damping term approaches zero, thus affecting the total score of the entire path. This results in severe penalties, causing any construction plan involving dangerous cross-operations to score extremely low and be naturally eliminated.

[0052] The entropy-reducing decision tree evolution engine traverses each major branch of the decision tree in parallel, calculating the cumulative performance of each path. To avoid invalid computations, the system also employs a forward-looking regret value strategy for dead-branch circuit breaking. When the system has computed halfway along a path, it records the cumulative performance already achieved. Simultaneously, it invokes a linear programming algorithm to quickly estimate the maximum potential energy that can be released by all remaining unconstructed elements along the path under the most ideal, unconstrained theoretical state. If found The value is already less than the score of the currently found globally optimal path. For example, a certain ratio If this happens, the system determines that no matter how much optimization is done, this path can never become the optimal solution. This can be understood as a dead end with increasing entropy, which should be immediately shut down to save massive amounts of computing resources and concentrate computing power on exploring more promising branches.

[0053] Step 5: Extract the dynamic scheduling priority index and generate instructions.

[0054] In this embodiment, after pruning and melting, the system obtains a set of high-quality paths with the highest scores. These paths represent several equally excellent and selectable construction schemes for the near future. The task of this embodiment is to extract the most important task to be performed at the current moment from these schemes.

[0055] The system introduces the concept of Dynamic Scheduling Priority Index (DSPI). For each BIM spatiotemporal element to be executed at the current moment... The system calculates its DSPI value using the following two metrics.

[0056] The first indicator is the frequency of occurrence. The system will collect statistics on volume elements. How many times did this occur within the first time step of the set of all retained high-quality paths? If 90 out of 100 high-quality paths recommend starting the task immediately... If so, its frequency of occurrence is 90%. This represents a high degree of consensus within the algorithm.

[0057] The second indicator is the delay gradient. The system will perform a simulation: assuming that the volume element is forcibly... The start time of construction is delayed by one time step, and then the total score of the scheduling function for the path is recalculated. The difference between the new score and the original score is the delay gradient. The larger the gradient, the more severe the damage to the global benefit from delaying the task; conversely, the smaller the gradient, the less urgent the task.

[0058] Finally, through the weighted formula Calculate the final priority index for each entity to be executed. Weighting coefficients. Adjustments will be made by the project manager based on their management style. Understandably, projects prioritizing stability can increase [their size / weight]. Prioritize tasks with high consensus levels based on their weight; projects aiming for maximum efficiency can increase the weight of tasks with high consensus levels. The weight of each task determines the priority of executing the task that has the greatest impact on the overall project duration.

[0059] After calculating the DSPI values ​​of all pending execution elements, the system sorts them in descending order of index. The task with the highest ranking is the scheduling instruction that should be executed at the current moment. These instructions are encapsulated into a visual electronic work order, containing information such as task name, component ID, 3D location diagram, and required resource list. Then, through a closed-loop feedback and instruction terminal module, it is pushed to the tablets or mobile apps of relevant personnel such as the on-site construction team leader, tower crane operator, and material handler.

[0060] Step Six: Implement a closed-loop feedback system at the site.

[0061] In order to ensure that the scheduling system is not a theoretical tool divorced from reality, a real-time, closed-loop feedback mechanism is crucial in this embodiment.

[0062] The system will periodically (e.g., every 4 hours) activate the sensing devices on site. Fixed laser scanners deployed at different locations in the convention center structure, or panoramic cameras carried by drones, will automatically collect high-precision point cloud or continuous image data of the construction site.

[0063] After receiving this massive amount of raw data, the closed-loop feedback and command terminal module invokes point cloud processing and computer vision algorithms to quickly reconstruct a realistic 3D model reflecting the current state of the construction site. Subsequently, the system performs crucial geometric registration and discrepancy comparison operations. It precisely aligns this newly generated realistic model with the BIM design model within the system on the same coordinate system.

[0064] By comparing the two models, the system can automatically identify which BIM spatiotemporal elements already exist in reality, meaning that these components have been completed. For completed elements, the system immediately updates their status: it resets their potential energy value in the construction potential energy field to zero, because the potential energy they contain has been "released"; at the same time, it unlocks the topological dependencies of all their immediate successor processes in the CPM network diagram, making these successor processes candidate tasks for the next round of scheduling.

[0065] If the system detects during the difference comparison that the actual construction progress in a certain area is significantly behind the expected progress of the scheduling instructions, it will identify that area as a "lagging area". In the next round, which restarts a few hours later, the system will dynamically and specifically increase the resource consumption of all unconstructed elements in the denominator of the scheduling function within the lagging area. This embodiment, based on a negative feedback mechanism, can automatically direct the system's attention to the lagging components, increasing their scheduling urgency and prompting resources and decisions to be tilted towards that area, thereby achieving dynamic correction of the project progress.

[0066] Thus, from BIM model analysis and potential field construction to constraint identification, decision tree evolution, function pruning, and finally instruction generation and closed-loop feedback, this embodiment comprehensively describes a dynamic scheduling method for the entire lifecycle of engineering projects that is adaptive, self-optimizing, and self-correcting. This method treats an engineering project as a complex system far from equilibrium, achieving globally optimal scheduling under multiple constraints and uncertainties by maximizing its spatiotemporal entropy reduction rate.

[0067] The above are merely preferred embodiments of the present invention and are not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A dynamic scheduling method for the entire lifecycle of an engineering project, characterized in that, The steps of this method include: Access the project management database, load and read the project BIM model data, discretize the geometric components into BIM spatiotemporal elements, assign a construction potential energy value to each BIM spatiotemporal element, and construct a construction potential energy field in three-dimensional space. Identify the gravity support constraints and work surface space constraints that the BIM spatiotemporal elements are subject to during construction, and construct a multi-dimensional hybrid constraint boundary by combining the inventory of human, machine and material resources in the project management database. Using the current site status as the root node, within the multidimensional hybrid constraint boundary, the construction sequence of the BIM spatiotemporal elements is subjected to discrete state fission to generate a multi-branch evolutionary decision tree. A scheduling function is defined with the optimization objective of maximizing the spatiotemporal entropy reduction rate of the engineering system. This function is used as the pruning operator of the multi-branch evolutionary decision tree. The cumulative efficiency of each branch path is calculated in the future time window, and dead branches are broken for branches with cumulative efficiency below a set threshold. From the set of retained optimal paths, extract the dynamic scheduling priority index of each BIM spatiotemporal element to be executed at the current moment, generate instructions to drive the on-site construction resource allocation, and generate scheduling instructions to drive on-site operations.

2. The dynamic scheduling method for the entire lifecycle of engineering projects according to claim 1, characterized in that, The construction potential energy field in the three-dimensional space is specifically constructed as follows: Extract the volume data and material properties of each geometric component from the engineering BIM model data, and calculate its engineering quantity weight; Based on the critical path method network graph, the spatiotemporal topological distance of each geometric component from the project delivery node is calculated, and the reciprocal of the spatiotemporal topological distance is defined as the urgency coefficient. Based on the first set formula, each BIM spatiotemporal element is assigned a construction potential energy value to construct a construction potential energy field in three-dimensional space.

3. The dynamic scheduling method for the entire lifecycle of engineering projects according to claim 2, characterized in that, The first set formula, its specific calculation formula is as follows: in, Let be the construction potential energy value released by the i-th BIM spatiotemporal element. As the weight of the project quantity, For spatiotemporal topological distance, and This is the preset adjustment weighting factor.

4. The dynamic scheduling method for the entire lifecycle of engineering projects according to claim 3, characterized in that, The construction of the multidimensional hybrid constraint boundary includes resource uncertainty correction based on Bayesian inference, specifically as follows: Retrieve resource supply logs from historical projects in the project management database, analyze the time deviation distribution of various resources from order placement to arrival, and establish a prior probability model. Furthermore, by using IoT gates and positioning tags deployed on-site, the actual entry records of the current batch of resources are collected as observational evidence. ; Based on Bayes' theorem Calculate the posterior probability distribution to obtain the confidence level of resource availability within a specific time window; Prior probability model With observational evidence The posterior probability distribution of resource supply delay is merged and updated, and the cumulative probability value of the posterior probability distribution within the preset safety interval is calculated as the availability confidence. An inverse proportional mapping function is established to transform the availability confidence into an elastic relaxation factor with a value in the range (0,1), so that the multi-branch evolutionary decision tree can shrink the branch search range during periods of risk in resource supply.

5. The dynamic scheduling method for the entire lifecycle of engineering projects according to claim 4, characterized in that, The generation of the multi-branch evolutionary decision tree specifically incorporates a Nash equilibrium game mechanism, which is as follows: When performing discrete state fission, identify multiple competing BIM spatiotemporal elements vying for the same scarce resource under the current node; Construct a multi-agent non-cooperative game matrix, using the construction potential energy value of each BIM spatiotemporal element. The revenue parameter is used, while the delay penalty caused by waiting is used as the cost parameter. To find the Nash equilibrium solution of the game matrix, only the BIM spatiotemporal volume element sorting combination that is in equilibrium is used as an effective branch to generate child nodes, and the others are discarded.

6. The dynamic scheduling method for the entire lifecycle of engineering projects according to claim 5, characterized in that, The defined scheduling function aims to maximize the spatiotemporal entropy reduction rate of the engineering system, and its specific calculation formula is as follows: in, For a branch path in a decision tree The cumulative spatiotemporal entropy reduction efficiency; k is the time step index, K is the total number of steps in the prediction time window; The set of BIM spatiotemporal volume elements executed in parallel within the k-th time step; The construction potential energy value released by the i-th BIM spatiotemporal element; This is the time sensitivity coefficient; For execution time; The current moment; Let be the resource consumption cost of the i-th individual element; This is the congestion penalty gain coefficient; Let be the set of volume element pairs that have spatial conflicts within the k-th time step; The bounding box overlap volume of the conflicting element pairs; This is the standard safe working space constant.

7. The dynamic scheduling method for the entire lifecycle of engineering projects according to claim 6, characterized in that, The The calculation steps are as follows: Read the geometric center coordinates and construction method type of the BIM spatiotemporal elements of the parallel operation; Based on the construction method type, assign the dynamic operation expansion radius to the BIM spatiotemporal elements to generate a three-dimensional dynamic bounding box. The intersection state of the three-dimensional dynamic bounding boxes of different BIM spatiotemporal elements is detected using the separation axis theorem. If they intersect, calculate the geometric volume of the intersecting portion based on Boolean intersection operations, and use it as... Substitute the values ​​into the scheduling function.

8. The dynamic scheduling method for the entire lifecycle of an engineering project according to claim 7, characterized in that, The specific steps for performing dead branch circuit breaking on branches with cumulative performance below a set threshold are as follows: Calculate the cumulative performance achieved by the current branch path; Estimate the maximum potential energy release value of the remaining unconstructed BIM spatiotemporal elements under theoretically unconstrained conditions. ; like Less than the currently found global optimal solution preset ratio If this branch path is considered a dead branch, the calculation of its subsequent nodes is immediately terminated. 。 9. The dynamic scheduling method for the entire lifecycle of engineering projects according to claim 8, characterized in that, The extraction of the dynamic scheduling priority index of each BIM spatiotemporal element to be executed at the current moment is specifically as follows: Calculate the frequency of BIM spatiotemporal elements to be executed in all valid paths that have not been broken. ; Calculate the delay gradient of BIM spatiotemporal elements ; The dynamic scheduling priority index is calculated using a second set formula, wherein the second set formula is: in, These are the weighting coefficients. It is a BIM spatiotemporal volume element.

10. A dynamic scheduling system for the entire lifecycle of an engineering project, characterized in that, include: The data parsing and potential energy field construction module connects to the engineering project management database, loads and reads the engineering BIM model data, discretizes the geometric components into BIM spatiotemporal elements, assigns a construction potential energy value to each BIM spatiotemporal element, and constructs a construction potential energy field in three-dimensional space. The constraint boundary and game theory calculation module identifies the gravity support constraints and work surface space constraints that the BIM spatiotemporal elements are subjected to during construction, and constructs a multi-dimensional hybrid constraint boundary by combining the inventory of human, machine and material in the project management database. The entropy reduction decision tree evolution engine takes the current engineering site state as the root node and performs discrete state fission on the construction sequence of the BIM spatiotemporal elements within the multidimensional hybrid constraint boundary to generate a multi-branch evolutionary decision tree. It defines a scheduling function with the optimization objective of maximizing the spatiotemporal entropy reduction rate of the engineering system as the pruning operator of the multi-branch evolutionary decision tree, traverses and calculates the cumulative efficiency of each branch path in the future time window, and performs dead branch melting on branches with cumulative efficiency lower than a set threshold. The closed-loop feedback and command terminal extracts the dynamic scheduling priority index of each BIM spatiotemporal element to be executed at the current moment from the set of retained optimal paths, generates commands to drive the on-site construction resource allocation, and generates scheduling commands to drive on-site operations.

Images