Cloud platform-based engineering project progress and resource collaborative management system

Abstract

This invention relates to the field of engineering project management technology and discloses a cloud-based engineering project progress and resource collaboration management system. The system generates a master progress sequence by collecting progress data and generates a resource flow record set by recording resource information in real time. The system extracts key progress nodes and corresponding resource consumption peak characteristics and performs time alignment verification to generate preliminary collaboration deviation information. The system inputs this information into a collaboration simulation model to perform parallel simulations of subsequent progress and resource requirements, obtaining a future collaboration simulation scenario. Based on this scenario, the system analyzes on-site image data to obtain actual collaboration state parameters and generates management conclusions for adjusting subsequent plans and resource allocation by comparing them with the simulation scenario. This invention achieves dynamic and accurate diagnosis and forward-looking closed-loop control of progress and resource collaboration status.

Description

Technical Field

[0001] This invention relates to the field of engineering project management technology, specifically to an engineering project progress and resource collaboration management system based on a cloud platform. Background Technology

[0002] Existing project management systems typically treat progress tracking and resource management as separate modules. Progress management records task deadlines and completion status, while resource management compiles data on the input and consumption of various resources. The two modules only perform periodic data aggregation and plan comparisons, lacking real-time analysis of the dynamic correlation between progress events and resource consumption at a micro-time scale.

[0003] Existing technical solutions have shortcomings. The system cannot automatically identify the degree of matching between the occurrence time of key progress nodes and the corresponding peak resource consumption. The peak of resource input may lag behind or precede the critical moment of actual project progress. This timing misalignment is a direct manifestation of coordination failure, but conventional systems only focus on whether the total amount is exceeded and cannot capture such time-series-based coordination deviations.

[0004] Existing systems rely heavily on static planning or simple extrapolation from single historical trends to predict project futures. Predictive models fail to incorporate identified schedule and resource coordination deviations as key input parameters, and lack a feedback mechanism for continuous verification and correction of predictions using real-time field data. This leads to a disconnect between forward-looking judgments used for management decisions and the dynamic process of actual project execution.

[0005] There is a need for a system that can perform time alignment checks on schedules and resources to diagnose micro-level collaborative status, and can integrate real-time data with historical deviations to perform closed-loop simulations, so as to achieve accurate positioning and forward-looking dynamic control of collaborative problems in engineering projects. Summary of the Invention

[0006] The purpose of this invention is to provide a cloud-based engineering project progress and resource collaborative management system to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides a cloud-based engineering project progress and resource collaboration management system, the system comprising:

[0008] The data acquisition module collects the actual progress execution data sequence of the target project and generates the main progress sequence of the project.

[0009] The record generation module records the resource allocation and consumption information of the target project in real time and generates a set of resource flow records;

[0010] The deviation verification module extracts several key progress nodes based on the main sequence of the project progress within a preset time window, extracts the resource consumption peak features corresponding to the key progress nodes based on the resource flow record set, and performs time alignment verification between the key progress nodes and the resource consumption peak features to generate preliminary collaborative deviation information.

[0011] The simulation module inputs the main sequence of the project schedule, the set of resource flow records, and the preliminary coordination deviation information into the pre-built collaborative simulation model to simulate the schedule events and resource requirements of subsequent project nodes in parallel, and obtain a collaborative simulation picture of the future time series.

[0012] The status analysis module extracts features and determines the status of the current site image data of the target project based on the collaborative projection picture of the future time sequence, and obtains the actual collaborative status parameters on site.

[0013] The management decision-making module compares the actual on-site collaborative status parameters with the collaborative projection map of the future time sequence, and generates collaborative management conclusions for adjusting the subsequent execution plan and resource allocation instructions of the project.

[0014] Preferably, the actual progress execution data sequence of the target project is collected to generate the main project progress sequence, including:

[0015] During the execution cycle of the target project, quantitative indicators reflecting the completion status of sub-projects are periodically collected according to the preset monitoring time interval. The quantitative indicators include the percentage of completed work, the acceptance mark of key processes, and the number of active on-site work units.

[0016] The quantitative indicators collected at each monitoring time interval are arranged and combined in chronological order to form a multidimensional data sequence containing timestamps and progress indicators.

[0017] Linear interpolation is used to fill discontinuous monitoring point data in the multidimensional data sequence, and smoothing filtering is applied to abnormally fluctuating index values ​​to generate a master sequence of project progress with continuity and consistency.

[0018] Preferably, the resource allocation and consumption information of the target project is recorded in real time to generate a resource flow record set, including:

[0019] Establish data connection channels with the target project's material warehouse, equipment sensors, and manual attendance terminals to receive real-time records of material inbound and outbound flows, equipment operating energy consumption and working hours, as well as the working hours invested by human resources for each type of work.

[0020] Each resource record is assigned an associated engineering structure decomposition code and construction task code, and resource data from different sources and in different formats are uniformly converted into a standardized resource event description format.

[0021] Standardized resource events are categorized by resource type, and within each resource type, a time series subset reflecting resource stock, increment, and consumption is constructed using time as the dimension. The time series subsets of all resource types are integrated to form a resource flow record set.

[0022] Preferably, within a preset time window, several key progress nodes are extracted based on the main sequence of the project schedule, and resource consumption peak features corresponding to the key progress nodes are extracted based on the resource flow record set. The key progress nodes and resource consumption peak features are then time-aligned to generate preliminary coordination deviation information, including:

[0023] Within the set backtracking time window, scan the main sequence of the project schedule, identify the time points when the rate of change in schedule exceeds the set threshold, and mark the time points as key progress nodes.

[0024] In the resource flow record set, locate the resource consumption record that is the same as the engineering structure decomposition code associated with each key progress node, and draw the consumption rate curve of various resources under the engineering structure decomposition code within the time window;

[0025] From each resource consumption rate curve, identify the time point corresponding to the local maximum value of the consumption rate, and define the time point and the corresponding resource category together as the resource consumption peak feature;

[0026] For each critical progress node, calculate the absolute difference between it and the peak characteristics of all associated resource consumption on the time axis. If the difference exceeds the preset allowable time tolerance, it is determined that the critical progress node has a coordination deviation. Summarize the coordination deviation determination results of all nodes to generate preliminary coordination deviation information.

[0027] Preferably, the main project schedule sequence, resource flow record set, and preliminary coordination deviation information are input into a pre-built collaborative simulation model to perform parallel simulations of the schedule events and resource requirements of subsequent project nodes, thereby obtaining a collaborative simulation scenario of the future time series, including:

[0028] The collaborative simulation model consists of a schedule evolution network and a resource coupling network connected in parallel. The schedule evolution network is used to receive the main schedule sequence of the project and the initial collaborative deviation information, while the resource coupling network is used to receive the set of resource flow records.

[0029] The progress evolution network infers the probability distribution of possible progress states on multiple future time slices based on the patterns of historical progress sequences and the current coordination deviations.

[0030] Based on historical resource flow patterns and the probability distribution of progress states output by the progress evolution network, the resource coupling network simulates the types, quantities, and arrival time windows of resources required to support each possible progress state, forming a cluster of resource demand scenarios.

[0031] By spatiotemporally matching and fusing the probability distribution of progress status with the cluster of resource demand scenarios, invalid scenarios with logical contradictions or resource conflicts are eliminated, and finally a collaborative projection picture of future time sequence containing time, progress events, resource events and their relationships is generated.

[0032] Preferred methods for constructing collaborative inference models include:

[0033] Collect complete historical data of a large number of completed similar projects, including historical progress master sequence, historical resource transfer records and historical collaboration problem records;

[0034] Extract the sequential constraints between schedule events, the mapping relationship between resource consumption and schedule events, and the triggering conditions and propagation paths of coordination deviation events from historical data;

[0035] Based on the aforementioned extracted relationships, mappings, and paths, a progress evolution network is constructed using a directed graph structure, where nodes represent progress events and edges represent constraints and evolutionary relationships between events.

[0036] A resource coupling network is constructed using resource categories as nodes and resource supply and demand events as edges. The connection weight of the resource coupling network is determined by the coupling strength between historical resource consumption and progress.

[0037] By iteratively training and optimizing the parameters in the progress evolution network and resource coupling network using historical data, the network can reproduce the historical collaborative deduction process until the simulation results output by the model match the actual historical development to a preset standard.

[0038] Preferably, based on the collaborative projection scenario of future time series, feature extraction and state determination are performed on the current on-site image data of the target project to obtain the actual on-site collaborative state parameters, including:

[0039] From the collaborative projection of future timelines, extract one or more key time segments that are about to arrive, as well as the core progress events and resource availability events that are expected to occur within the time segments;

[0040] The image acquisition equipment deployed at the corresponding work site of the target project is scheduled to collect a set of real-time images reflecting the completion of core progress events and the availability of resources.

[0041] Using an image recognition network trained for engineering scenarios, real-time image sets are processed to identify the assembly status of specific engineering components, the working posture of construction machinery, and the inventory characteristics of material storage yards in the images.

[0042] The identified states and features are compared with the expected events of the corresponding time segments in the collaborative simulation scenario. The completion evidence of progress events and the matching evidence of resource events are quantified and calculated. The two types of evidence are combined to generate actual collaborative state parameters on site.

[0043] Preferably, an image recognition network trained for engineering scenarios is used to process real-time image sets to identify the assembly status of specific engineering components, the working posture of construction machinery, and the inventory characteristics of material storage yards, including:

[0044] The image recognition network adopts a multi-branch feature extraction architecture, in which one branch is dedicated to identifying and locating predefined engineering component types in the image and outputting their geometric dimensions, installation angles, and relative positional relationships as assembly status features.

[0045] Another branch is used to identify the type of construction machinery, the posture of the working device, and its interaction with the surrounding environment, encoding the posture information into feature vectors as working posture features;

[0046] The third branch is used to segment the material yard area, estimate the volume or floor space of different material categories in the yard, compare it with the reference template, and output the richness characteristics of the inventory.

[0047] The feature vectors extracted from the three branches are fused and input into a state classifier, which outputs a judgment description of the overall collaborative operation state reflected in the current image.

[0048] Preferably, the actual on-site collaborative status parameters are compared with the collaborative projection diagram of future time series to generate collaborative management conclusions for adjusting subsequent project execution plans and resource allocation instructions, including:

[0049] The progress completion evidence and resource matching evidence reflected in the actual on-site collaborative status parameters are compared item by item with the expected values ​​of the corresponding nodes in the collaborative simulation scenario, and the difference vector is calculated.

[0050] Analyze the direction and magnitude of the difference vector. If the difference indicates that the schedule is ahead of schedule or resources are sufficient, then find the subsequent links in the collaborative simulation scenario that can compress the schedule or optimize resource input.

[0051] If the discrepancy indicates a delay in progress or a shortage of resources, an impact propagation analysis is performed in the collaborative simulation scenario to simulate the chain and magnitude of the impact of the deviation on all subsequent dependent nodes and resource plans.

[0052] Based on the current deviation, impact analysis results, and the preset scheduling rule base, a series of specific plan adjustment suggestions and resource allocation instructions are generated. The plan adjustment suggestions and resource allocation instructions together constitute the collaborative management conclusion.

[0053] Preferably, the generation of planned adjustment suggestion items and resource allocation instruction items is based on a preset scheduling rule base, and the method for constructing the scheduling rule base includes:

[0054] The project management field summarizes expert experience, standard construction process specifications, and historical successful scheduling cases, and transforms them into formalized production rules, each of which includes preconditions and conclusion actions.

[0055] The preconditions describe the specific type, degree, and project context of the coordination deviation, and the conclusion action describes the schedule change measures or resource reallocation solutions that should be taken to deal with such deviations;

[0056] The formalized production rules are stored in a knowledge graph, and the rules are connected through engineering logic relationships.

[0057] When collaborative management conclusions need to be generated, the difference vector and impact analysis results are used as input to the knowledge graph to trigger matching rules. The conclusion actions are then instantiated into specific plan adjustment suggestions and resource allocation instructions.

[0058] Compared with the prior art, the beneficial effects of the present invention are:

[0059] Coordination deviation identification based on time alignment verification can extract key nodes from the progress data sequence and locate corresponding consumption peaks in resource flow data. By directly comparing the correspondence between the two on the time axis, the system achieves accurate diagnosis of whether resource input and key process execution are synchronized. This technical solution can detect in real time progress delays caused by resource allocation delays or micro-level disconnections caused by ineffective resource accumulation that fail to effectively promote key nodes. This allows the identification of coordination problems to move from macro-level total monitoring to the micro-level time sequence matching, providing specific time and event targets for precise intervention.

[0060] The co-simulation model incorporates schedule sequences containing timing deviation information, resource records, and preliminary coordination deviations, driving parallel simulations of subsequent schedule events and resource requirements. This simulation process embeds identified coordination misalignment patterns, enabling the generated future co-simulation scenario to reflect multiple potential paths for the propagation and evolution of existing problems. This ensures that the prediction results are not merely a simple extension of the ideal plan, but rather include the inheritance and extrapolation of historical coordination defects, enhancing the prediction results' ability to warn of potential risks and characterize real-world complexity.

[0061] Guided by the simulated scenario, feature extraction and state determination are performed on the on-site images to obtain actual collaborative state parameters, which are then compared in real time with the simulated scenario. This process constitutes a dynamic closed loop of "simulation prediction - on-site verification." This mechanism can continuously calibrate the output of the simulation model and correct the simulation path based on the real-time on-site status. Ultimately, management decisions are generated based on the collaborative scenario dynamically calibrated from on-site data, enabling resource allocation and schedule adjustment instructions to respond to actual fluctuations in project execution, forming an adaptive feedback control mechanism that improves the real-time nature and situational adaptability of decision-making. Attached Figure Description

[0062] Figure 1 This is a sequence diagram of the cloud-based engineering project progress and resource collaborative management system described in this invention.

[0063] Figure 2 A flowchart for generating the master sequence of project schedules;

[0064] Figure 3 A flowchart for generating preliminary coordination deviation information;

[0065] Figure 4 A trend chart showing the consumption rate of various types of resources during the resource transfer phase of an engineering project;

[0066] Figure 5 This is a comparison chart of resource consumption rate and availability rate during the resource status analysis phase of an engineering project. Detailed Implementation

[0067] 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. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0068] Please see Figure 1This invention provides a cloud-based engineering project progress and resource collaborative management system. The system includes: a data acquisition module that collects the actual progress execution data sequence of the target engineering project and generates a master progress sequence; a record generation module that records the resource configuration and consumption information of the target engineering project in real time and generates a resource flow record set; a deviation verification module that, within a preset time window, extracts several key progress nodes based on the master progress sequence, extracts resource consumption peak features corresponding to the key progress nodes based on the resource flow record set, and performs time alignment verification between the key progress nodes and the resource consumption peak features to generate preliminary collaborative deviation information; a simulation module that inputs the master progress sequence, resource flow record set, and preliminary collaborative deviation information into a pre-constructed collaborative simulation model to perform parallel simulation of the progress events and resource requirements of subsequent engineering nodes, obtaining a collaborative simulation scenario for the future time series; and a state analysis module that, based on the collaborative simulation scenario for the future time series, extracts features and determines the state of the current site image data of the target engineering project to obtain actual on-site collaborative state parameters. The management decision module compares the actual on-site collaborative status parameters with the collaborative projection map of the future time sequence, and generates collaborative management conclusions for adjusting the subsequent execution plan and resource allocation instructions of the project.

[0069] In one embodiment of the present invention, see [reference] Figure 2 Within the execution cycle of the target project, quantitative indicators reflecting the completion status of sub-projects are periodically collected according to preset monitoring time intervals. These quantitative indicators include the percentage of completed work, key process acceptance marks, and the number of active on-site work units. The quantitative indicators collected at each monitoring time interval are arranged and combined in chronological order to form a multi-dimensional data sequence containing timestamps and progress indicators. Linear interpolation is used to fill discontinuous monitoring point data in the multi-dimensional data sequence, and abnormally fluctuating indicator values ​​are smoothed and filtered to generate a main project progress sequence with continuity and consistency. Data connection channels are established with the target project's material warehouse, equipment sensors, and manual attendance terminals to receive real-time records of material inbound and outbound flows, equipment operating energy consumption and working hours, and the working hours of human resources invested by various trades. Each resource record is appended with its associated engineering structure decomposition code and construction task code, and resource data from different sources and in different formats are uniformly converted into a standardized resource event description format. Standardized resource events are categorized by resource type, and within each resource type, a time series subset reflecting resource stock, increment, and consumption is constructed using time as the dimension. The time series subsets of all resource types are integrated to form a resource flow record set.

[0070] In practical implementation, for a high-rise building construction project, the data acquisition module collects quantitative indicators reflecting the completion status of sub-projects on a daily basis according to a preset monitoring time interval. In practice, these quantitative indicators include the percentage of completed basement floor concrete pouring, the acceptance marks of key steel structure hoisting processes, and the number of active on-site work units for curtain wall installation. The quantitative indicators collected at each monitoring time interval are arranged and combined chronologically to form a multi-dimensional data sequence containing timestamps and progress indicators. In some embodiments, the timestamps are accurate to the day, and the progress indicators are stored in numerical or binary format. Linear interpolation is used to fill discontinuous monitoring point data in the multi-dimensional data sequence. In practice, if the "number of active on-site work units" indicator for a particular day is missing due to equipment failure, the arithmetic mean of the values ​​for the two days before and after is used to fill the gap. Smoothing filters are applied to abnormally fluctuating indicator values ​​to generate a continuous and consistent master sequence of project progress. This smoothing process employs a five-day sliding window, applying a weighted moving average algorithm to the "percentage of completed work" indicator data within the window to suppress the impact of abnormal daily fluctuations. The smoothing formula can be expressed as:

[0071]

[0072] in: Represents a point in time Smoothed progress indicator values Represents the original sequence at time point The progress indicator value, Represents the offset The weight coefficients, and satisfying .

[0073] Establish data connection channels with the target project's material warehouse, equipment sensors, and manual attendance terminals. In specific implementation, the data connection channels connect to the RFID entry and exit scanning terminals for materials such as steel bars and concrete via standard IoT protocols, to the operating status sensors of tower cranes and concrete pump trucks via equipment communication interfaces, and to the mobile attendance terminals of laborers via a wireless network. Real-time records of material entry and exit flows, equipment operating energy consumption and working hours, and the working hours of various trades are received. In some embodiments, the material entry flow records include material codes, specifications, quantities, entry times, and storage warehouse location information; the equipment operating energy consumption and working hour records include equipment numbers, start and end times, and fuel or electricity consumption data.

[0074] Each resource record is associated with a structural decomposition code and a construction task code. In practice, a batch of steel reinforcement outbound records will be accompanied by the structural decomposition code for the planned construction of "main structure - third-floor beams and slabs" and the construction task code for "steel reinforcement binding". Furthermore, resource data from different sources and in different formats are uniformly converted into a standardized resource event description format. This standardized resource event description format specifies a unified timestamp format, resource measurement unit, and uses a structured template of "event type: resource category, quantity, associated code, time point" for description.

[0075] Standardized resource events are categorized by resource type. In specific implementations, resource categories include "reinforcing steel," "concrete," "formwork," "tower crane shifts," and "reinforcing steel worker hours." Within each resource category, a time-series subset reflecting resource inventory, increment, and consumption is constructed. In specific implementations, for the "concrete" resource category, each data point in the time-series subset includes the date, the cumulative amount received that day, the cumulative amount consumed that day, and the remaining inventory calculated based on the previous day's inventory. All time-series subsets of all resource categories are integrated to form a resource flow record set. In some embodiments, this resource flow record set is stored in a multi-dimensional data table in the cloud platform database, with each row representing the inventory, increment, and consumption status of a resource category at a given point in time.

[0076] In one embodiment of the present invention, see [reference] Figure 3 Within a set backtracking time window, the main sequence of the project schedule is scanned to identify the points in time where the rate of change in schedule exceeds a set threshold, and these points are marked as critical nodes in the schedule. In the resource flow record set, resource consumption records with the same engineering structure decomposition code associated with each critical node are located, and consumption rate curves for various resources under the engineering structure decomposition code are plotted within the time window. From each resource consumption rate curve, the time point corresponding to the local maximum consumption rate is identified, and this time point and the corresponding resource category are defined together as resource consumption peak features. For each critical node, the absolute difference between it and all associated resource consumption peak features on the time axis is calculated. If the difference exceeds a preset allowable time tolerance, the critical node is determined to have a coordination deviation. The coordination deviation judgment results of all nodes are summarized to generate preliminary coordination deviation information.

[0077] In practice, the deviation verification module performs a collaborative status assessment during one construction phase of the main structure of a high-rise building. The module sets a backtracking time window of 10 working days, starting from the 50th working day of the project schedule. Within the set backtracking time window, the module scans the main sequence of the project schedule and identifies the points in time when the progress change rate exceeds a set threshold. In practice, the progress change rate is calculated based on the "percentage of completed concrete pouring area per floor" indicator, with a set threshold of more than 5% per day. The module identifies that on the 55th working day, this indicator increases by 12% in a single day, and on the 58th working day, it increases by 8% in a single day. These points in time are marked as critical progress nodes. In some embodiments, the marked critical progress nodes include the time points "T55" and "T58", and are associated with their corresponding structural decomposition code "S-02-03" (representing the third floor of the second construction section).

[0078] Within the resource flow record set, the module locates resource consumption records with the same engineering structure decomposition code associated with each critical progress node. In specific implementation, the module filters all resource consumption events with the engineering structure decomposition code "S-02-03" from the resource flow record set. It then plots the consumption rate curves of various resources under the aforementioned engineering structure decomposition code within a time window. For example, for the resource category "ready-mixed concrete," the module calculates its daily outbound consumption from the 50th to the 59th working day and plots a consumption rate curve with the date on the horizontal axis and the daily consumption on the vertical axis. For the resource category "reinforcing steel worker hours," the module calculates its total daily working hours and plots another consumption rate curve. From each resource consumption rate curve, the module identifies the time point corresponding to the local maximum consumption rate. In specific implementation, by finding points on the curve where the first derivative is zero and the second derivative is negative, the module identifies that the "ready-mixed concrete" consumption rate curve reaches a local maximum on the 56th working day, and the "reinforcing steel worker hours" consumption rate curve reaches a local maximum on the 57th working day. The time point and the corresponding resource category are jointly defined as resource consumption peak characteristics. In some embodiments, the defined resource consumption peak characteristics include "ready-mixed concrete peak: T56" and "steel reinforcement worker hour peak: T57".

[0079] For each critical schedule node, the absolute difference between it and all associated resource consumption peaks on the time axis is calculated. In specific implementation, for the critical schedule node "T55", the absolute difference between it and "ready-mixed concrete peak: T56" is calculated to be 1 day, and the absolute difference between it and "reinforcing steel worker peak: T57" is calculated to be 2 days. If the difference exceeds the preset allowable time tolerance, the critical schedule node is determined to have a coordination deviation. It can be understood that the allowable time tolerance is set to 1 day. Therefore, the 2-day difference between the critical schedule node "T55" and "reinforcing steel worker peak: T57" exceeds the allowable time tolerance and is determined to have a coordination deviation. However, the time difference between the critical schedule node "T58" and all associated peaks does not exceed the allowable time tolerance. The results of coordination deviation assessments for all nodes are summarized to generate preliminary coordination deviation information. In practice, this information is presented as a structured list containing the following entries: "Node T55 (S-02-03 level progress jump) has a 2-day delay deviation from the peak consumption of the critical resource 'rebar worker hours'; no significant time deviation was detected at node T58." Deviation identification is based on the progress change rate, calculated using the following formula:

[0080]

[0081] in: Represents a point in time The absolute change in the progress indicator. Representing a point in time The progress indicator value, Represents the previous point in time The progress indicator value, when Greater than the preset threshold At that time, point in time It was identified as a critical milestone in the schedule.

[0082] In one embodiment of the present invention, the collaborative simulation model is composed of a schedule evolution network and a resource coupling network connected in parallel. The schedule evolution network receives the main schedule sequence of the project and preliminary coordination deviation information, while the resource coupling network receives a set of resource flow records. Based on the patterns of historical schedule sequences and the current coordination deviation, the schedule evolution network deduces the probability distribution of possible schedule states on multiple future time slices. Based on historical resource flow patterns and the schedule state probability distribution output by the schedule evolution network, the resource coupling network simulates the types, quantities, and arrival time windows of resources required to support each possible schedule state, forming a cluster of resource demand scenarios. The schedule state probability distribution and the resource demand scenario cluster are spatiotemporally matched and fused, eliminating invalid scenarios with logical contradictions or resource conflicts, ultimately integrating them to generate a collaborative simulation scenario of future time sequences containing time, schedule events, resource events, and their relationships.

[0083] The method for constructing the collaborative simulation model includes: collecting complete historical data from a large number of completed similar projects, including historical progress master sequences, historical resource flow records, and historical collaborative problem records. From the historical data, the sequential constraints between progress events, the mapping relationship between resource consumption and progress events, and the triggering conditions and propagation paths of collaborative deviation events are extracted. Based on the extracted relationships, mappings, and paths, a directed graph structure is used to construct a progress evolution network, where nodes represent progress events and edges represent constraints and evolutionary relationships between events. A resource coupling network is constructed using resource categories as nodes and resource supply and demand events as edges. The connection weights of the resource coupling network are determined by the coupling strength between historical resource consumption and progress. The parameters in the progress evolution network and the resource coupling network are iteratively trained and optimized using historical data, enabling the network to reproduce the historical collaborative simulation process until the model output simulation results achieve a preset standard of consistency with the actual historical development.

[0084] In practical implementation, the simulation module simulates the future collaborative status of a high-rise building project in the main structural construction phase. The collaborative simulation model consists of a schedule evolution network and a resource coupling network connected in parallel. The schedule evolution network receives the main project schedule sequence up to the current date and preliminary collaborative deviation information regarding the delay of rebar worker resources. The resource coupling network receives a set of resource flow records within the same time period. Based on the patterns of historical schedule sequences and the current collaborative deviation, the schedule evolution network infers the probability distribution of possible schedule states on multiple future time slices. In practical implementation, based on patterns learned from historical data, the schedule evolution network takes into account the current deviation information of "two-day delay in rebar worker hours" and outputs the predicted probability of the "floor completion status" for each day of the next seven days. For example, the probability of completing the concrete pouring of the third floor slab on the second day is predicted to be 85%, but the probability of completing the rebar binding of the fourth floor wall is adjusted to 40% due to resource delays.

[0085] Based on historical resource flow patterns and the probability distribution of progress states output by the progress evolution network, the resource coupling network simulates the types, quantities, and arrival time windows of resources required to support each possible progress state, forming a cluster of resource demand scenarios. For example, for the high-probability state predicted by the progress evolution network as "completing the concrete pouring of the third floor slab on the second day," the resource coupling network simulates a resource requirement of 80 cubic meters of ready-mixed concrete, 20 man-days of formwork labor, and resources in place before 9:00 AM on the second day. For the low-probability state of "completing the rebar binding of the fourth floor wall," the simulation indicates a requirement of 15 man-days of rebar labor and 5 tons of rebar, with the demand time window adjusted to the third or fourth day due to delays. The probability distribution of progress states is spatiotemporally matched and fused with the cluster of resource demand scenarios, eliminating invalid scenarios with logical contradictions or resource conflicts. In some embodiments, the module identifies an invalid scenario where the simulated progress event of "constructing the fifth floor on the fifth day" relies on a tower crane resource that is completely occupied within the same time window by the demand of another higher-probability progress event; this conflicting scenario is therefore eliminated. Ultimately, a collaborative projection map of future time series is generated, which includes time, progress events, resource events and their relationships. In specific implementation, the collaborative projection map of future time series is presented as a directed graph with time series and probability weights. Nodes represent different progress events and resource events from day two to day seven, and edges represent the dependencies and triggering relationships between events.

[0086] The method for constructing the collaborative simulation model includes: collecting complete historical data from a large number of completed similar projects, including historical progress master sequences, historical resource flow records, and historical collaborative problem records. In specific implementation, the collected historical data covers 20 completed high-rise residential projects. The data for each project includes daily progress records from foundation construction to main structure completion, consumption records of various labor, materials, and machinery, as well as logs of collaborative problems caused by design changes, material shortages, and severe weather. From the historical data, the sequential constraints between progress events, the mapping relationship between resource consumption and progress events, and the triggering conditions and propagation paths of collaborative deviation events are extracted. Optionally, the extracted rules include that the "floor concrete pouring" progress event can only begin after the completion of the "formwork erection" and "reinforcement binding" progress events, and that the "peak consumption of ready-mixed concrete" usually occurs on the day the "floor concrete pouring" progress event begins.

[0087] Based on the extracted relationships, mappings, and paths, a directed graph structure is used to construct a progress evolution network. Nodes represent progress events, and edges represent constraints and evolutionary relationships between events. In some embodiments, nodes include "rebar binding completed," "formwork erection completed," and "concrete pouring started," while edges are labeled with minimum time intervals and logical conditions. A resource coupling network is constructed using resource categories as nodes and resource supply and demand events as edges. The connection weights of this resource coupling network are determined by the coupling strength between historical resource consumption and progress. For example, the connection weight between the "rebar worker hours" node and the "rebar binding completed" progress node is determined by the correlation coefficient and regression coefficient between the historical rebar binding workload and rebar worker hours. The parameters in the progress evolution network and the resource coupling network are iteratively trained and optimized using historical data, enabling the network to reproduce the historical collaborative deduction process until the model's output simulation results match the historical actual development to a preset standard. In specific implementation, the training process minimizes the differences between the simulated progress sequence and simulated resource consumption curve derived by the model and the historical real sequence and curve. The optimization process continues until the matching index exceeds 95%. The objective function for model optimization can be expressed as minimizing the overall discrepancy between historical data and simulation results:

[0088]

[0089] in: This represents the overall loss value; the smaller the value, the higher the degree of agreement. Represents a historical timeline. Represents the progress sequence of the model simulation; Represents the historical resource consumption curve. The resource consumption curve represents the model simulation; The function represents the calculation of the root mean square error between two sets of data; and It is a hyperparameter used to balance the weights of schedule error and resource error.

[0090] In one embodiment of the present invention, one or more upcoming key time segments, along with core progress events and resource availability events expected to occur within the time segments, are extracted from a future time-series collaborative projection scenario. Image acquisition devices deployed at the corresponding work surfaces of the target engineering project site are scheduled to collect a set of real-time images reflecting signs of completion of core progress events and resource availability. An image recognition network trained for the engineering scenario is used to process the real-time image set, identifying the assembly status of specific engineering components, the working posture of construction machinery, and the inventory characteristics of material storage areas. The identified states and features are compared with the expected events of the corresponding time segments in the collaborative projection scenario, quantifying the completion evidence of progress events and the matching evidence of resource events, and combining the two types of evidence to generate actual collaborative state parameters on site.

[0091] The image recognition network employs a multi-branch feature extraction architecture. One branch is dedicated to identifying and locating predefined engineering component types in the image, outputting their geometric dimensions, installation angles, and relative positional relationships as assembly state features. Another branch identifies the type of construction machinery, the posture of the working device, and its interaction with the surrounding environment, encoding posture information into feature vectors as working posture features. A third branch segments the material storage area, estimates the volume or floor space of different material categories within the storage area, compares it with a reference template, and outputs inventory richness features. The feature vectors extracted from the three branches are fused and input into a state classifier, which outputs a judgment description of the overall collaborative operation state reflected in the current image.

[0092] In practical implementation, the state analysis module determines the on-site collaborative state of a standard floor construction cycle of a high-rise building project based on the collaborative projection scenario of future time series. From the collaborative projection scenario of future time series, it extracts one or more upcoming key time segments, as well as the core progress events and resource placement events expected to occur within the time segments. In practical implementation, the module extracts the next 48 hours as the key time segment. The core progress events expected to occur within this segment include "completion of shear wall reinforcement binding on the east side of S-05 floor" and "completion of formwork installation on the west side of S-05 floor". The expected resource placement events include "HRB400 rebar (25mm specification) arrives at the east side working face" and "large steel formwork (number MB-05) is placed on the west side working face".

[0093] The module automatically schedules image acquisition equipment deployed at the corresponding work surfaces of the target project site. In specific implementation, the module automatically schedules a fixed panoramic camera installed on the east side of the fifth-floor construction surface and a track-mounted inspection camera on the west side to collect a set of real-time images reflecting the completion signs of core progress events and the availability of resources. In some embodiments, the real-time image set includes three high-definition images from different angles on the east work surface and keyframe screenshots of a 30-second video stream from the west work surface. An image recognition network trained for the engineering scenario is used to process the real-time image set, identifying the assembly status of specific engineering components, the working posture of construction machinery, and the inventory characteristics of material storage areas. It can be understood that the image recognition network is loaded with pre-trained weights for engineering elements such as steel reinforcement mesh, formwork systems, tower cranes, and material stacks.

[0094] The image recognition network employs a multi-branch feature extraction architecture. One branch is dedicated to identifying and locating predefined engineering component types in the image, outputting their geometric dimensions, installation angles, and relative positional relationships as assembly status features. In specific implementations, for images of the eastern working face, this branch identifies the spacing of vertical reinforcing bars, the number of horizontal reinforcing bar tying points, and the overall coverage area of ​​the reinforcing bar mesh, outputting three numerical features: "average reinforcing bar spacing," "tying point density," and "mesh coverage." Another branch identifies the type of construction machinery, the posture of the working device, and its interaction with the surrounding environment, encoding the posture information into feature vectors as working posture features. Optionally, for images of the western working face, this branch identifies the presence of a tower crane in the image and calculates its boom elevation angle and the planar coordinates of its hook relative to the formwork stacking area. The third branch is used to segment the material yard area, estimate the volume or area occupied by different material categories in the yard, compare it with the reference template, and output the richness feature of the inventory. In some embodiments, this branch performs pixel-level segmentation of the material yard in the edge area of ​​the eastern image, identifies the outline of the rebar bundles and estimates the approximate volume of their stack, and outputs the "rebar stack volume ratio" feature.

[0095] The feature vectors extracted from the three branches are fused and input into a state classifier. The state classifier outputs a judgment description of the overall collaborative operation status reflected by the current image. It can be understood that the fused feature vector includes assembly status features, work posture features, and inventory richness features. Based on these fused features, the state classifier outputs a multi-dimensional state vector. The identified states and features are compared with the expected events of the corresponding time segments in the collaborative simulation scenario to quantify the completion evidence of progress events and the matching evidence of resource events. In specific implementation, the module compares the "rebar mesh coverage rate" output by the image recognition network with the expected completion threshold of the event "rebar binding of the east shear wall of S-05 floor completed" in the collaborative simulation scenario, and calculates the progress completion evidence value as 0.85; it compares the "rebar stack volume ratio" feature with the expected inventory of the expected event "HRB400 rebar arrives at the east working face", and calculates the resource matching evidence value as 0.90. Combining the two types of evidence generates actual on-site collaborative status parameters. In some embodiments, these parameters are a structure containing a timestamp, work surface location, progress completion evidence value, resource matching evidence value, and a comprehensive status code. The matching evidence can be calculated using the following formula:

[0096]

[0097] in: The value represents the evidence of a good match. This represents the number of feature dimensions used for comparison. Representing the Preset weights for each feature dimension in the comparison; The function represents the calculation of recognition features. With expected characteristics A function of similarity between features, for example, for numerical features, the normalized difference complement can be used; Represents the first extracted from the image One eigenvalue; The first expected scenario in the collaborative simulation picture Each feature value. See Table 1 for feature matching comparison.

[0098] Table 1: Comparison Table of Image Recognition Features and Expected Events

[0099]

[0100] See Figure 4This is a trend chart showing the consumption rates of various resources during the resource flow phase of an engineering project. The peak consumption of rebar and steel formwork shows a time lag, closely matching the construction process of "rebar tying → formwork installation." Rebar consumption has the highest peak, making it the core resource input in this phase. This type of chart is used for analyzing consumption patterns during the resource flow phase, matching resource consumption with construction procedures, verifying the rationality of resource input, and enabling advance planning for the procurement and storage of high-consumption resources. It also optimizes equipment time allocation and improves resource utilization efficiency.

[0101] In one embodiment of the present invention, the progress completion evidence and resource matching evidence reflected in the actual on-site collaborative status parameters are compared item by item with the expected values ​​of the corresponding nodes in the collaborative simulation scenario to calculate the difference vector. The direction and magnitude of the difference vector are analyzed. If the difference indicates that the schedule is ahead of schedule or resources are sufficient, the collaborative simulation scenario is used to identify subsequent stages where the schedule can be compressed or resource input can be optimized. If the difference indicates that the schedule is behind schedule or resources are scarce, an impact propagation analysis is performed in the collaborative simulation scenario to simulate the impact chain and magnitude of the deviation on all subsequent dependent nodes and resource plans. By combining the current deviation, the impact analysis results, and the preset scheduling rule base, a series of specific plan adjustment suggestions and resource allocation instructions are generated. These plan adjustment suggestions and resource allocation instructions together constitute the collaborative management conclusion.

[0102] The generation of planned adjustment suggestions and resource allocation instructions is based on a pre-defined scheduling rule base. The construction method of this rule base includes: summarizing expert experience, standard construction process specifications, and historical successful scheduling cases in the field of engineering project management, and transforming them into formalized production rules. Each rule includes preconditions and a conclusion action. The preconditions describe the type, degree, and engineering context of the coordination deviation, while the conclusion action describes the schedule change measures or resource reallocation schemes to be taken to address such deviations. These formalized production rules are stored in a knowledge graph, and the rules are connected through engineering logic relationships. When a collaborative management conclusion needs to be generated, the difference vector and impact analysis results are used as query inputs to the knowledge graph, triggering a matching rule. The conclusion action, after instantiation, becomes the specific planned adjustment suggestion and resource allocation instruction.

[0103] In practice, the management decision-making module responds to the rebar tying operation in the main structure construction of a high-rise building based on the actual collaborative state parameters output by the state analysis module. It compares the progress completion evidence and resource matching evidence reflected in the actual collaborative state parameters with the expected values ​​of the corresponding nodes in the collaborative simulation scenario, calculating the difference vector. In the specific implementation, for the key node "rebar tying of the east shear wall of floor S-05," the progress completion evidence value in the actual collaborative state parameters is 0.6, while the expected progress completion evidence value for this node in the collaborative simulation scenario is 0.9; the resource matching evidence value in the actual collaborative state parameters is 1.0, while the expected resource matching evidence value for this node in the collaborative simulation scenario is 0.9. The difference vector calculated by the module is [-0.3, +0.1], where negative values ​​represent progress delays and positive values ​​represent resource sufficiency better than expected.

[0104] The module analyzes the direction and magnitude of the discrepancy vector. If the discrepancy indicates that the schedule is ahead of schedule or resources are sufficient, it searches for subsequent stages with compressible timelines or optimized resource inputs in the collaborative simulation scenario. In some embodiments, the discrepancy vector shows that the schedule is lagging but resources are sufficient, so the module does not perform the operation of searching for stages with compressible timelines, but instead proceeds to the lag analysis process. If the discrepancy indicates that the schedule is lagging or resources are scarce, it performs an impact propagation analysis in the collaborative simulation scenario, simulating the impact chain and magnitude of the deviation on all subsequent dependent nodes and resource plans. It can be understood that the module takes the currently lagging "S-05 floor east side shear wall reinforcement binding" node as the starting point, traverses all subsequent dependent schedule nodes in the directed graph of the collaborative simulation scenario, including "S-05 floor east side formwork closure", "S-05 floor concrete pouring", "S-06 floor construction preparation", etc., and recursively calculates the delay in the start time of each subsequent node that may be caused by the current 0.3 unit of schedule lag evidence value, based on the process logic relationship and minimum time interval defined in the scenario, forming an impact propagation path and a list of its corresponding time delay magnitudes.

[0105] Based on the current deviation, impact analysis results, and a pre-defined scheduling rule base, a series of specific plan adjustment suggestions and resource allocation instructions are generated. These suggestions and instructions together constitute the collaborative management conclusion. In practice, the current deviation is "the reinforcement binding progress of the shear wall on the east side of floor S-05 is lagging by 30%," and the impact analysis results show that "the concrete pouring node of floor S-05 is expected to be delayed by 8 hours," and "the construction preparation node of floor S-06 is expected to be delayed by 5 hours." The management decision module inputs this information, along with information on current resource sufficiency, into the scheduling rule base for matching and querying.

[0106] The generation of planned adjustment suggestions and resource allocation instructions is based on a pre-defined scheduling rule library. The construction method of this scheduling rule library includes: summarizing expert experience, standard construction process specifications, and historical successful scheduling cases in the field of engineering project management, and transforming them into formalized production rules. Each rule includes a precondition and a conclusion action. Optionally, a typical formalized production rule can be expressed as: "Precondition: The rebar tying progress is lagging by more than 20% and the associated concrete pouring node has not yet started, and there are sufficient rebar workers on site; Conclusion Action: Instruction 1, immediately dispatch 2 more rebar workers to the east work face; Instruction 2, it is recommended to postpone the start time of subsequent formwork closure work by 4 hours." The preconditions describe the type, degree, and project context of the coordination deviation, while the conclusion action describes the schedule change measures or resource reallocation schemes to be taken to address such deviations.

[0107] The formalized production rules are stored in a knowledge graph, and the rules are connected through engineering logic relationships. In some embodiments, the nodes of the knowledge graph represent different construction procedures, resource types, or deviation states, and the edges represent triggering, substitution, or mutual exclusion relationships between rules. When it is necessary to generate collaborative management conclusions, the difference vector and the impact analysis results are used as query inputs to the knowledge graph to trigger matching rules. The conclusion actions are instantiated to become specific plan adjustment suggestion items and resource allocation instruction items. It can be understood that, for the aforementioned example, after inputting the query, the knowledge graph retrieves and triggers matching rules, instantiating "add 2 steelworkers" in the rule conclusion actions as "dispatch two steelworkers, Zhang XX and Li XX, from the west reserve team to the east work surface, and execute immediately," and instantiating "delay by 4 hours" as "adjust the planned start time of the S-05 floor east side formwork closure operation from the original 08:00 the next day to 12:00 the next day." The formula for calculating the total delay time in the impact propagation analysis can be expressed as:

[0108]

[0109] in: This represents the total equivalent delay time caused by the impact on the propagation chain starting from the current deviation node; This represents the set of all subsequent dependent nodes in the collaborative simulation scenario that are directly affected by the current deviation; Representing the The weight coefficient of each subsequent node in the propagation path is determined by the criticality of the process. The current deviation is expected to lead to the first The delay in the start time of each subsequent node.

[0110] See Figure 5This is a chart comparing resource consumption and availability during the resource status analysis phase of an engineering project. Machinery and steel should be prioritized for replenishment to avoid construction interruptions; for cement and human resources, reserve strategies can be adjusted to optimize costs. The significant differences in the matching degree between consumption and availability rates for different resources reflect an imbalance in resource allocation. This type of chart is used for supply and demand balance assessment during the resource status analysis phase. It helps to quickly identify resource shortages and promptly initiate replenishment processes; optimize the reserve levels of sufficient resources to reduce inventory costs; and adjust resource allocation strategies to achieve a balance between supply and demand for various resources.

[0111] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0112] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A cloud-based engineering project progress and resource collaboration management system, characterized in that, The system includes: The data acquisition module collects the actual progress execution data sequence of the target project and generates the main progress sequence of the project. The record generation module records the resource configuration and consumption information of the target project in real time and generates a resource flow record set, which is stored in the cloud platform database in the form of a multi-dimensional data table. The deviation verification module, within a preset time window, extracts several key progress nodes based on the main sequence of the project schedule, extracts resource consumption peak features corresponding to the key progress nodes based on the resource flow record set, and verifies the time alignment between the key progress nodes and the resource consumption peak features to generate preliminary coordination deviation information, specifically including: Scan the main timeline of the project schedule and mark the points where the rate of change in schedule exceeds a set threshold as critical milestones in the schedule. Locate the resource consumption records that have the same engineering structure decomposition code as each critical progress node, and plot the consumption rate curves of various resources under the engineering structure decomposition code within the time window. From each resource consumption rate curve, identify the time point corresponding to the local maximum value of the consumption rate, and define the time point and the corresponding resource category together as the resource consumption peak feature; Calculate the absolute difference between each critical progress node and the peak characteristics of all associated resource consumption on the time axis. If the difference exceeds the preset allowable time tolerance, it is determined that the critical progress node has a coordination deviation. Summarize the coordination deviation determination results of all nodes to generate preliminary coordination deviation information. The simulation module inputs the main project schedule sequence, resource flow record set, and preliminary coordination deviation information into a pre-built collaborative simulation model. It then performs parallel simulations of the schedule events and resource requirements of subsequent project nodes to obtain a collaborative simulation scenario for the future timeline, specifically including: The collaborative simulation model includes a schedule evolution network with schedule events as nodes and constraints and evolutionary relationships between events as edges, and a resource coupling network with resource categories as nodes and resource supply and demand events as edges. The progress evolution network infers the probability distribution of possible progress states on multiple future time slices based on the patterns of historical progress sequences and the current coordination deviations. Based on historical resource flow patterns and the probability distribution of progress states output by the progress evolution network, the resource coupling network simulates the types, quantities, and arrival time windows of resources required to support each possible progress state, forming a cluster of resource demand scenarios. The probability distribution of progress status is spatiotemporally matched and fused with the cluster of resource demand scenarios. Invalid scenarios with logical contradictions or resource conflicts are eliminated, and finally a collaborative inference picture of future time sequence containing time, progress events, resource events and their correlation is generated. The status analysis module extracts features and determines the status of the current site image data of the target project based on the collaborative projection picture of the future time sequence, and obtains the actual collaborative status parameters on site. The management decision-making module compares the actual on-site collaborative status parameters with the collaborative projection map of the future time sequence, and generates collaborative management conclusions for adjusting the subsequent execution plan and resource allocation instructions of the project.

2. The cloud-based engineering project progress and resource collaboration management system according to claim 1, characterized in that, The actual progress execution data sequence of the target project is collected to generate the main project progress sequence, including: During the execution cycle of the target project, quantitative indicators reflecting the completion status of sub-projects are periodically collected according to the preset monitoring time interval. The quantitative indicators include the percentage of completed work, the acceptance mark of key processes, and the number of active on-site work units. The quantitative indicators collected at each monitoring time interval are arranged and combined in chronological order to form a multidimensional data sequence containing timestamps and progress indicators. Linear interpolation is used to fill discontinuous monitoring point data in the multidimensional data sequence, and smoothing filtering is applied to abnormally fluctuating index values ​​to generate a master sequence of project progress with continuity and consistency.

3. The cloud-based engineering project progress and resource collaboration management system according to claim 1, characterized in that, Real-time recording of resource allocation and consumption information for the target project generates a resource flow record set, including: Establish data connection channels with the target project's material warehouse, equipment sensors, and manual attendance terminals to receive real-time records of material inbound and outbound flows, equipment operating energy consumption and working hours, as well as the working hours invested by human resources for each type of work. Each resource record is assigned an associated engineering structure decomposition code and construction task code, and resource data from different sources and in different formats are uniformly converted into a standardized resource event description format. Standardized resource events are categorized by resource type, and within each resource type, a time series subset reflecting resource stock, increment, and consumption is constructed using time as the dimension. The time series subsets of all resource types are integrated to form a resource flow record set.

4. The cloud-based engineering project progress and resource collaboration management system according to claim 1, characterized in that, Other methods for constructing collaborative inference models include: Collect complete historical data of a large number of completed similar projects, including historical progress master sequence, historical resource transfer records and historical collaboration problem records; Extract the sequential constraints between schedule events, the mapping relationship between resource consumption and schedule events, and the triggering conditions and propagation paths of coordination deviation events from historical data; The connection weight of the resource coupling network is determined by the coupling strength between historical resource consumption and progress. By iteratively training and optimizing the parameters in the progress evolution network and resource coupling network using historical data, the network can reproduce the historical collaborative deduction process until the simulation results output by the model match the actual historical development to a preset standard.

5. The cloud-based engineering project progress and resource collaboration management system according to claim 1, characterized in that, Based on the collaborative projection scenario of future time series, feature extraction and state determination are performed on the current on-site image data of the target project to obtain the actual on-site collaborative state parameters, including: From the collaborative projection of future timelines, extract one or more key time segments that are about to arrive, as well as the core progress events and resource availability events that are expected to occur within the time segments; The image acquisition equipment deployed at the corresponding work site of the target project is scheduled to collect a set of real-time images reflecting the completion signs of core progress events and the availability of resources. Using an image recognition network trained for engineering scenarios, real-time image sets are processed to identify the assembly status features of specific engineering components, the working posture features of construction machinery, and the inventory features of material stockpiles in the images. The identified features are compared with the expected events of the corresponding time segments in the collaborative simulation scenario. The progress completion rate evidence and resource matching rate evidence of the progress events are quantitatively calculated. The two types of evidence are combined to generate the actual collaborative state parameters on site.

6. The cloud-based engineering project progress and resource collaborative management system according to claim 5, characterized in that, Using an image recognition network trained for engineering scenarios, a real-time image dataset is processed to identify the assembly status features of specific engineering components, the working posture features of construction machinery, and the inventory features of material storage yards, including: The image recognition network adopts a multi-branch feature extraction architecture, in which one branch is dedicated to identifying and locating predefined engineering component types in the image and outputting their geometric dimensions, installation angles, and relative positional relationships as assembly status features. Another branch is used to identify the type of construction machinery, the posture of the working device, and the interaction with the surrounding environment, encoding the posture information into feature vectors as the working posture features of the construction machinery. The third branch is used to segment the material yard area, estimate the volume or floor space of different material categories in the yard, compare it with the reference template, and output the inventory characteristics of the material yard. The feature vectors extracted from the three branches are fused and input into a state classifier, which outputs a judgment description of the overall collaborative operation state reflected in the current image.

7. The cloud-based engineering project progress and resource collaborative management system according to claim 1, characterized in that, By comparing the actual on-site collaborative status parameters with the collaborative projection diagrams for future timelines, collaborative management conclusions are generated to adjust subsequent project execution plans and resource allocation instructions, including: The progress completion evidence and resource matching evidence reflected in the actual on-site collaborative status parameters are compared item by item with the expected values ​​of the corresponding nodes in the collaborative simulation scenario, and the difference vector is calculated. Analyze the direction and magnitude of the difference vector. If the difference vector indicates that the schedule is ahead of schedule or resources are sufficient, then find the subsequent links in the collaborative simulation scenario that can compress the schedule or optimize resource input. If the discrepancy vector indicates a schedule lag or resource shortage, then an impact propagation analysis is performed in the co-simulation scenario to simulate the chain and magnitude of the impact of the schedule or resource deviation represented by the discrepancy vector on all subsequent dependent nodes and resource plans. By combining the difference vector, the impact analysis results, and the preset scheduling rule base, a series of specific plan adjustment suggestions and resource allocation instructions are generated. The plan adjustment suggestions and resource allocation instructions together constitute the collaborative management conclusion.

8. The cloud-based engineering project progress and resource collaborative management system according to claim 7, characterized in that, The generation of planned adjustment suggestion items and resource allocation instruction items is based on a preset scheduling rule base, and the method for constructing the scheduling rule base includes: The project management field summarizes expert experience, standard construction process specifications, and historical successful scheduling cases, and transforms them into formalized production rules, each of which includes preconditions and conclusion actions. The preconditions describe the specific type, degree, and project context of the coordination deviation, and the conclusion action describes the schedule change measures or resource reallocation schemes that should be taken to deal with this type of coordination deviation. The formalized production rules are stored in a knowledge graph, and the rules are connected through engineering logic relationships. When collaborative management conclusions need to be generated, the difference vector and impact analysis results are used as input to the knowledge graph to trigger matching rules. The conclusion actions are then instantiated into specific plan adjustment suggestions and resource allocation instructions.

Images