Intelligent monitoring method and system for engineering project progress

Abstract

The present invention relates to the technical field of engineering project management, and discloses a method and system for intelligent monitoring of engineering project progress. The method comprises: obtaining progress parameters including task completion status, resource consumption, etc. in real time through a multi-source data acquisition terminal; inputting the parameters into a preset anomaly detection model to identify and mark abnormal data segments, and the model dynamically adjusts the threshold according to historical data characteristics; then inputting the marked parameters into a collaborative prediction model to generate progress trend prediction results, and the model performs multi-dimensional fitting based on task dependencies and resource allocation weights; finally, using a priority scheduling algorithm to classify and integrate the prediction results and current progress parameters to form a dynamic monitoring data set for storage. The present invention realizes intelligent monitoring of engineering project progress, improves monitoring accuracy and timeliness, can effectively prevent project delays and cost overruns, and improves project management efficiency.

Description

Technical Field

[0001] The present invention relates to the technical field of engineering project management, and in particular to an engineering project progress intelligent monitoring method and system. Background Art

[0002] In the field of engineering project management, effective project progress monitoring is crucial to ensuring on-time and high-quality project delivery. Traditional methods of project progress monitoring have many drawbacks and are unable to meet the growing complexity and refined management requirements of modern engineering projects.

[0003] In the early days, project progress monitoring relied primarily on regular manual inspections and record-keeping. Workers frequently visited construction sites to manually record data such as work completed and resource usage. This method was not only labor-intensive and time-consuming, inefficient, but also susceptible to human error, making data accuracy difficult to guarantee. For example, in large-scale construction projects, where construction areas are extensive, manual record-keeping could result in omissions or errors, leading to skewed assessments of project progress.

[0004] With the development of information technology, some engineering projects have begun to rely on simple tools such as spreadsheets to assist with progress management. While this represents a certain improvement over manual record-keeping, this method still requires manual data entry and organization, making it impossible to achieve real-time dynamic monitoring. As project scale expands and data volumes increase dramatically, problems such as delayed data updates and difficulty integrating information become more prominent. For example, in complex municipal engineering projects involving multiple construction units and numerous stages, data from different construction areas is difficult to aggregate and analyze in real time, making it difficult for managers to keep abreast of the overall project progress.

[0005] Some engineering projects have introduced some basic project management software with simple progress tracking functions, such as Gantt chart drawing. However, these software often lack the ability to deeply integrate and analyze multi-source data. They can usually only process a single type of data and are unable to comprehensively consider multiple factors such as task completion status, resource consumption, construction period deviation rate, and quality inspection results to fully evaluate project progress. Moreover, the prediction functions of these software are mostly based on simple linear extrapolations, and do not fully consider the complex dependencies between tasks and resource allocation weights, resulting in poor prediction results and an inability to provide a reliable basis for project decision-making. In large-scale water conservancy projects, the various construction links are closely interrelated and resource allocation is complex. Basic project management software is difficult to accurately predict project progress, which may put the project at risk of delays.

[0006] Furthermore, existing technologies also have shortcomings when it comes to handling abnormal data. When project schedule anomalies occur, traditional methods often fail to quickly and accurately identify them, making it even more difficult to take targeted measures based on the type and severity of the anomaly. Manually troubleshooting anomalies in the face of massive amounts of project data is both time-consuming and labor-intensive, and can easily miss critical information. For example, in rail transit construction projects, if schedule anomalies occur, relying on manual analysis to identify the cause can delay the optimal time to resolve the issue, increasing project costs. Summary of the Invention

[0007] The purpose of the present invention is to provide a method and system for intelligently monitoring the progress of an engineering project to solve the problems raised in the above background technology.

[0008] To achieve the above-mentioned object, the present invention provides the following technical solution: a method for intelligently monitoring the progress of an engineering project, the method comprising:

[0009] Acquire project progress parameters in real time through multi-source data acquisition terminals, including task completion status, resource consumption, construction period deviation rate, and quality inspection results;

[0010] Inputting the progress parameter into a preset anomaly detection model to identify and mark abnormal data segments, wherein the anomaly detection model dynamically adjusts the detection threshold based on the characteristics of different anomaly types in the historical engineering data;

[0011] Input the marked progress parameters into a preset collaborative prediction model to generate a progress trend prediction result, wherein the collaborative prediction model performs multi-dimensional fitting based on task dependencies and resource allocation weights;

[0012] The prediction results and current progress parameters are classified and integrated using a preset priority scheduling algorithm to form a dynamic monitoring data set and store it.

[0013] Preferably, the step of constructing the anomaly detection model includes: obtaining a historical engineering data set, wherein each data in the historical engineering data set is marked with an anomaly type and an impact level; dividing training subsets based on the anomaly type and impact level, each training subset corresponding to an abnormal scenario; using the training subsets to train the initial detection model in parallel, until the recognition accuracy of the initial detection model for each abnormal scenario is greater than or equal to a preset first threshold, the training is stopped to obtain an intermediate detection model; the historical engineering data set is input into the intermediate detection model to verify whether the anomaly recognition result output by the intermediate detection model meets the preset error range; if so, the intermediate detection model is determined as the anomaly detection model.

[0014] Preferably, the real-time acquisition of the progress parameters of the engineering project through the multi-source data acquisition terminal includes:

[0015] Establishing a communication connection with a target monitoring terminal, wherein the target monitoring terminal is deployed at a preset monitoring node of the engineering project;

[0016] Continuously reading the real-time data of the target monitoring terminal according to a preset collection period, and marking the collection time point based on the time series characteristics of the real-time data;

[0017] According to the task topology relationship of the engineering project, the real-time data of different monitoring nodes at the same time point are logically aligned to form a set of associated progress parameters.

[0018] Preferably, inputting the progress parameter into a preset anomaly detection model comprises:

[0019] Extracting a mutation data segment from the progress parameter, wherein the mutation data segment is a data segment in which the parameter change amplitude exceeds a preset mutation threshold within a continuous monitoring period;

[0020] Generate an anomaly assessment index based on the duration and degree of deviation of the mutation data segment;

[0021] The corresponding detection algorithm is dynamically selected according to the anomaly assessment index, wherein the isolation forest algorithm is used for short-term high-amplitude mutations and the cluster analysis algorithm is used for long-term low-amplitude mutations.

[0022] Preferably, the method further comprises:

[0023] After marking the abnormal data segment, performing data consistency check on the progress parameter;

[0024] If the verification finds that the data conflict rate exceeds a preset second threshold, the collaborative prediction model is triggered to correct the priority of the conflicting data, wherein the high-priority conflicting data is the task data segment that affects the critical path.

[0025] Preferably, the collaborative prediction model includes the following prediction steps:

[0026] Constructing a dynamic weight model based on the task dependency network of the engineering project, wherein each task node corresponds to a dependency weight coefficient;

[0027] Calculate the collaborative prediction weight based on the resource consumption difference of adjacent task nodes;

[0028] Combined with the historical trends of the progress parameters, multi-dimensional prediction and completion of missing task nodes are performed.

[0029] Preferably, the method further comprises:

[0030] After the prediction is completed, the prediction results are logically verified, where the verification method includes comparing the execution deviation between the prediction data and the actual task nodes;

[0031] If the deviation exceeds a preset third threshold, the collaborative prediction weight is readjusted and the prediction is iterated until the deviation is less than the third threshold.

[0032] Preferably, classifying and integrating the prediction results and current progress parameters using a preset priority scheduling algorithm includes:

[0033] Classify the first-level classification labels according to the task type, wherein the first-level classification labels include critical tasks, non-critical tasks, and risky tasks;

[0034] Under each level of classification label, the second level classification sub-label is further divided based on the urgency of the task;

[0035] The classified task data is stored in different partitions of the time series database according to the label level.

[0036] Preferably, the method further comprises:

[0037] Configure access levels for classification tags based on pre-set task permissions;

[0038] Upon receiving a data retrieval request, verify whether the permission identifier provided by the requester matches the access level of the target classification label;

[0039] If there is a match, the data retrieval channel for the corresponding classification label is opened.

[0040] Preferably, the present invention further includes an intelligent monitoring system for project progress, the system comprising:

[0041] A multi-source data acquisition module is used to obtain the progress parameters of the engineering project in real time through a multi-source data acquisition terminal, wherein the progress parameters include task completion status, resource consumption, construction period deviation rate and quality inspection results;

[0042] an anomaly detection module, configured to input the progress parameter into a preset anomaly detection model to identify and mark abnormal data segments, wherein the anomaly detection model dynamically adjusts the detection threshold based on the characteristics of different anomaly types in the historical engineering data;

[0043] A collaborative prediction module is used to input the marked progress parameters into a preset collaborative prediction model to generate a progress trend prediction result, wherein the collaborative prediction model performs multi-dimensional fitting based on task dependencies and resource allocation weights;

[0044] Priority scheduling is used to classify and integrate the prediction results and current progress parameters using a preset priority scheduling algorithm to form a dynamic monitoring data set and store it.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] At the data collection and integration level, multi-source data collection terminals are used to obtain project progress parameters in real time, covering key information such as task completion status, resource consumption, construction period deviation rate, and quality inspection results. Multi-source data collection terminals can automatically and continuously collect data, greatly improving the efficiency and accuracy of data collection and avoiding the subjectivity and errors of manual collection. By continuously reading data and marking the collection time points according to the preset collection cycle, and logically aligning data based on task topology relationships, the collected data forms an organic set of related progress parameters, fully presenting the actual progress of the project and providing a comprehensive and reliable data foundation for subsequent analysis. For example, in large-scale bridge construction projects, sensors at various construction sites collect data in real time, accurately reflecting the construction progress of each part of the bridge, material usage, etc., providing project managers with precise information.

[0047] The construction and application of the anomaly detection model is a highlight of the present invention. The model dynamically adjusts the detection threshold based on the characteristics of different anomaly types in historical engineering data. Compared with the traditional fixed threshold detection method, it can more accurately identify abnormal data segments. By dividing the training subsets and training the initial detection model in parallel, the model's recognition accuracy for various abnormal scenarios is improved. In practical applications, when the project progress is abnormal, the model can quickly and accurately mark the abnormal data, allowing project managers to discover potential problems in a timely manner. For example, in road construction projects, when resource consumption suddenly increases or the construction period deviates significantly, the anomaly detection model can quickly capture this abnormal information and buy time for taking corrective measures.

[0048] The collaborative forecasting model uses multi-dimensional fitting based on task dependencies and resource allocation weights to generate more accurate and reliable progress trend forecasts. The construction of a dynamic weighting model, the calculation of collaborative forecasting weights, and the multi-dimensional forecast completion of missing task nodes fully account for the various complex factors in the project. Logical verification and iterative forecasting after forecast completion further enhance forecast accuracy. This enables project managers to understand project progress trends in advance, plan resource allocation, adjust construction plans in advance, and effectively prevent delays. For example, in a water conservancy hub construction project, by accurately forecasting the progress of each construction link, human and material resources can be rationally allocated to ensure the project proceeds as planned.

[0049] A preset priority scheduling algorithm is used to categorize and integrate forecast results and current progress parameters, allowing project managers to quickly access critical information. First-level classification tags are assigned based on task type, and second-level sub-tags are assigned based on urgency. These tags are then stored in different partitions of the time series database, improving the efficiency of data query and analysis. Furthermore, access levels for classification tags are configured based on task permissions, ensuring data security and confidentiality. In large-scale commercial complex construction projects, personnel at different levels can only access data with corresponding permissions, ensuring that sensitive information is not leaked. Project managers can also quickly locate progress data for key tasks and make timely decisions.

[0050] Overall, the intelligent project progress monitoring method and system of the present invention effectively improves the accuracy, timeliness and intelligence level of project progress monitoring, can significantly reduce the risks of project delays and cost overruns, improve the overall quality and management efficiency of the project, and provide a strong guarantee for the smooth implementation of the project. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 This is a working principle diagram of the method for intelligent monitoring of engineering project progress according to the present invention;

[0052] Figure 2 Flowchart obtained for progress parameters;

[0053] Figure 3 Flowchart for abnormal data detection;

[0054] Figure 4 Flowchart of data access control. DETAILED DESCRIPTION

[0055] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0056] See also Figures 1-4 The present invention provides a method for intelligently monitoring the progress of engineering projects, aiming to achieve comprehensive, accurate, and intelligent monitoring of the progress of engineering projects. The specific steps are as follows:

[0057] Multi-source data acquisition terminals collect real-time project progress parameters, including task completion status, resource consumption, schedule deviation rates, and quality inspection results. Multi-source data acquisition terminals can be deployed at key locations across the project using a variety of hardware devices, including sensors and monitoring equipment, to ensure real-time and accurate data capture.

[0058] The acquired progress parameters are fed into a pre-set anomaly detection model. This model dynamically adjusts detection thresholds based on the characteristics of different anomaly types in historical project data to identify and flag anomalous data segments. This approach allows for timely detection of anomalies in project progress, providing a basis for subsequent action.

[0059] The marked progress parameters are input into the pre-set collaborative forecasting model. The collaborative forecasting model performs a multi-dimensional fit based on task dependencies and resource allocation weights to generate a progress trend forecast. This forecast helps project managers understand the project's progress trajectory in advance and make informed decisions.

[0060] Using a pre-set priority scheduling algorithm, progress trend forecasts and current progress parameters are categorized and integrated to form and store a dynamic monitoring data set. This stored data can be accessed at any time for project progress analysis and monitoring, allowing project managers to keep abreast of project progress.

[0061] The present invention will be further described below in conjunction with Examples 1 to 5:

[0062] Example 1: This example details the construction of an anomaly detection model in actual project monitoring. First, a historical engineering dataset is obtained. This dataset contains a large amount of data from past projects, and each data entry is annotated with the anomaly type and impact level. For example, anomaly types may include resource overconsumption, construction delays, and substandard quality. The impact level is divided into three levels: high, medium, and low, depending on the degree of impact of the anomaly on the overall progress and quality of the project.

[0063] Based on this labeled information, training subsets are created. For example, for a specific construction project, all data showing progress anomalies due to resource overconsumption are grouped into one training subset, corresponding to the resource overconsumption anomaly scenario. Data showing progress anomalies due to quality issues are grouped into another training subset, corresponding to the quality anomaly scenario. Similarly, each training subset corresponds to a specific anomaly scenario.

[0064] After dividing the training subsets, the initial detection model is trained in parallel using these subsets. During training, the model parameters are continuously adjusted until the initial detection model's recognition accuracy for each abnormal scenario is greater than or equal to a preset first threshold. Training is then terminated, resulting in the intermediate detection model. Assuming the preset first threshold is 90%, the training is considered effective when the model's recognition accuracy for various abnormal scenarios, such as resource overconsumption and quality anomalies, reaches 90% or higher.

[0065] Next, the historical engineering dataset is fed into the intermediate detection model to verify whether its output of anomaly identification results meets the preset error range. If so, this intermediate detection model is designated as the final anomaly detection model. During the verification process, the error is calculated by comparing the anomaly identification results output by the model with the anomaly information already annotated in the dataset. For example, if the model identifies 100 anomalies, 95 of which are consistent with the anomaly information, and the error is within the preset range, the intermediate detection model is considered to meet the requirements and can be put into use as the official anomaly detection model.

[0066] Example 2: In actual project progress monitoring scenarios, obtaining accurate and timely progress parameters is the basis for effective monitoring. Taking a large residential construction project as an example, the process of obtaining project progress parameters in real time through a multi-source data acquisition terminal is described in detail.

[0067] This residential construction project features numerous pre-set monitoring nodes meticulously placed at key construction sites. For example, targeted monitoring terminals were installed in each building's foundation construction area, main structure construction floors, pipeline laying points, and equipment installation locations. These terminals possess diverse data collection capabilities, capturing a wide range of real-time data, including construction worker numbers, material usage, equipment operating status, and completion times for each construction phase. These data serve as crucial foundations for determining progress parameters.

[0068] To achieve stable data transmission, the project team established a reliable communication network to ensure effective communication connections between multi-source data acquisition terminals and target monitoring terminals. Based on the complex environment of the construction site, a combination of wired and wireless networks was adopted. For relatively fixed areas with less interference, such as monitoring points inside the main building, wired network connections are preferred to ensure the stability and reliability of data transmission. For areas where construction locations frequently change or wiring is difficult, such as outdoor earthwork construction sites and temporary material storage points, wireless network connections, such as 4G / 5G communication technology or Wi-Fi, are used to enable flexible access to the data acquisition system for monitoring terminals.

[0069] Setting a reasonable preset collection cycle is crucial for timely data acquisition. After analyzing the project's construction process and data variation characteristics, a half-hour collection cycle was determined. At the beginning of each collection cycle, the multi-source data collection terminal automatically sends a data read instruction to each target monitoring terminal. Upon receiving the instruction, the target monitoring terminal quickly transmits the stored real-time data to the collection terminal. During the data transmission process, the collection terminal accurately marks the collection time point based on the timing characteristics of the real-time data. This process is achieved with the help of high-precision clock synchronization technology, ensuring that all collected data has accurate timestamps. For example, during one data collection operation, the collection terminal successfully read data from the concrete pouring construction monitoring terminal on the 5th floor of Building 3 at 9:30:00 am and marked the time as 9:30:00, providing an accurate time basis for subsequent data processing and analysis.

[0070] Real-time data collected comes from different monitoring nodes, each of which has complex logical relationships. Therefore, it's necessary to logically align the real-time data from different monitoring nodes at the same time based on the project's task topology. In residential construction projects, task topology reflects the sequence and interdependencies between various construction tasks. For example, in the construction of Building 3, foundation construction must be completed before main structure construction can begin, and pipeline laying and equipment installation can only begin after the main structure has reached a certain stage of construction. Data collected at 9:30 a.m. from the Building 3 foundation construction monitoring node indicates that concrete pouring has reached 80%, while data from the main structure construction monitoring node indicates that rebar tying is underway on the fifth floor, and the pipeline laying monitoring node is in a waiting state. By analyzing the task topology, this data is logically aligned, clarifying its position and interrelationship within the overall project progress, ultimately forming a set of associated progress parameters. This integration of previously scattered and isolated data into a coherent whole provides strong support for accurate subsequent analysis of the project's progress.

[0071] Example 3: In project progress monitoring, accurately identifying abnormal data is crucial for timely discovering potential problems and ensuring smooth project progress. The following uses a municipal road construction project as an example to explain the specific process of inputting progress parameters into a pre-set anomaly detection model.

[0072] During municipal road construction, progress parameters contain a wealth of information, such as the daily amount of pavement material laid, construction crew work hours, deviations between the actual and planned progress, and quality inspection results for each construction section. When analyzing these progress parameters, the first step is to extract abrupt data segments. Setting a preset abrupt threshold is key to identifying abrupt data segments. For example, for the parameter pavement material laying quantity, based on analysis of data from previous similar projects and consideration of actual project construction conditions, a data segment is considered abrupt if the variation in pavement material laying quantity exceeds 30% within two consecutive monitoring cycles (assuming each monitoring cycle is one day). During a particular week of construction, 100 tons of pavement material were laid on Monday, then suddenly increased to 140 tons on Tuesday. This represents a 40% variation, exceeding the preset abrupt threshold. Therefore, the pavement material laying quantity data for the period from Monday to Tuesday is considered abrupt.

[0073] For the identified mutation data segments, it is necessary to generate anomaly evaluation indicators based on their duration and degree of deviation. Taking the mutation data segment of pavement material laying as an example, assuming that its duration is sky( is an integer greater than or equal to 1), the degree of deviation is (Degree of deviation By calculating the difference between the current data and the historical average data and comparing it with the historical average data, the calculation formula for the abnormal evaluation index is set as: ,in represents the abnormal evaluation index, Indicates a reference time, which is set to 7 days in this project. For example, if the above road material paving amount mutation data segment lasts for 3 days and the deviation degree is 40%, then the abnormal evaluation index This anomaly assessment indicator can comprehensively reflect the degree of abnormality of the mutation data segment and provide a basis for the subsequent selection of appropriate detection algorithms.

[0074] Based on the generated anomaly assessment indicators, the corresponding detection algorithm is dynamically selected to further analyze the anomaly. When a short-term, high-amplitude sudden change occurs, the isolation forest algorithm is used. In municipal road construction projects, if construction workers' working hours suddenly increase significantly on a single day, far exceeding the normal range, and this situation only lasts for one or two days, this is considered a short-term, high-amplitude sudden change. The isolation forest algorithm effectively identifies such relatively isolated data points in the data space, known as anomalies. It constructs multiple binary trees and maps each data point to a path length within the tree for evaluation. The shorter the path length, the further away the data point is from the normal data distribution, and the more likely it is an anomaly. For example, on a certain construction section, construction workers normally work between 8 and 10 hours a day. However, on one day, the number suddenly increases to 15 hours. Using the isolation forest algorithm to analyze the construction workers' working hours during this period can quickly identify this data as an anomaly.

[0075] Cluster analysis algorithms are used when long-term, low-amplitude, sudden changes occur. For example, during road construction, if the quality inspection results of a construction section meet quality standards for an extended period (e.g., 10 consecutive days), but remain consistently lower than those of previous similar projects or other sections within the project, and the fluctuations are relatively small, this is considered a long-term, low-amplitude sudden change. Cluster analysis algorithms group similar data points into clusters and determine whether the data is abnormal by analyzing the differences between clusters and the position of each data point within a cluster. In this project, cluster analysis was performed on the quality inspection results of each construction section based on various indicators (such as smoothness and compaction). If the quality inspection data of a section consistently falls into a cluster that differs significantly from other sections, and the overall quality indicators of this cluster are low, it can be determined that the construction quality data of that section is abnormal, indicating potential construction problems that require further investigation and resolution. By selecting appropriate detection algorithms based on different sudden changes, abnormal data in progress parameters can be more accurately identified, providing strong support for project progress monitoring.

[0076] Example 4: This example provides a detailed description of the prediction steps of the collaborative prediction model. In a complex water conservancy project, a dynamic weight model is first constructed based on the task dependency network of the project. The task dependency network reflects the sequence and dependency relationship between tasks, and each task node corresponds to a dependency weight coefficient. For example, there is a dependency relationship between the dam construction task and the flood discharge facility construction task. The completion of the dam construction task will affect the implementation of the flood discharge facility construction task. Based on the degree of this dependency, dependency weight coefficients are set for the dam construction task node and the flood discharge facility construction task node respectively.

[0077] The collaborative prediction weight is calculated based on the resource consumption difference of adjacent task nodes. Assume that there are two adjacent task nodes A and B, and the resource consumption of task node A in a period of time is , the resource consumption of task node B is , the calculation formula of collaborative prediction weight is: ,in This formula can be used to calculate the reasonable collaborative prediction weight based on the difference in resource consumption.

[0078] Combined with the historical trends of progress parameters, multi-dimensional predictions are performed to complete missing task nodes. During the implementation of water conservancy projects, data for certain task nodes may be missing due to various reasons. For example, the progress data for a small ancillary facility construction task is missing for a certain period of time. In this case, the historical progress data trend of this task node is used, combined with information from other related task nodes and the calculated collaborative prediction weights, to perform predictions and completion from multiple dimensions such as time and resources to ensure data integrity and provide accurate data support for subsequent progress predictions.

[0079] After the prediction is complete, the prediction results are logically verified. This verification involves comparing the predicted data with the actual execution deviation of the task nodes. For example, suppose the predicted completion progress of a task node at a specific time point is 80%, while the actual completion progress is 70%. The deviation between the two is calculated to be 10%. If the deviation exceeds a preset third threshold (assuming 5%), the collaborative prediction weights are readjusted and the prediction is iterated until the deviation is less than the third threshold. Through continuous adjustment and iteration, the accuracy of the prediction results is improved.

[0080] Example 5: This example provides a detailed introduction to the classification and integration of prediction results and current progress parameters using a preset priority scheduling algorithm. In a large commercial complex construction project, the first-level classification labels are first divided according to the task type. Tasks that affect the critical path of the project, such as the main building structure construction tasks, are classified as critical tasks; auxiliary tasks that have less impact on the overall progress of the project, such as landscaping construction tasks, are classified as non-critical tasks; and tasks that have certain risks and may have an adverse impact on the project progress, such as intelligent system installation tasks involving the application of new technologies, are classified as risky tasks.

[0081] Under each level of classification, sub-tags are further divided based on the urgency of the task. For example, for the critical building structure construction task, the task that is about to enter a critical construction phase and cannot be delayed is assigned the Urgent sub-tag, while the less urgent subsequent construction phase is assigned the Normal sub-tag.

[0082] The classified task data is stored in different partitions of the time series database according to the label level. For example, the task data under the critical task-urgent sub-label is stored in a specific partition of the database, which is convenient for quick query and access.

[0083] In addition, the access level of the classification tag is configured according to the preset task permissions. In the commercial complex construction project, senior project managers have higher permissions and can access data of all classification tags; while ordinary construction workers can only access data of non-critical tasks with lower access levels. When receiving a data retrieval request, the system will verify whether the permission identifier provided by the requester matches the access level of the target classification tag. If it matches, the data retrieval channel for the corresponding classification tag will be opened. For example, when an ordinary construction worker requests to retrieve data of a landscape greening construction task (which belongs to the non-critical task category and has a lower access level), the system verifies his permission identifier and allows him to obtain the relevant data if it matches, thus ensuring the security and confidentiality of the data.

[0084] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0085] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent monitoring of project progress, characterized in that: include: Acquire project progress parameters in real time through multi-source data acquisition terminals, including task completion status, resource consumption, construction period deviation rate, and quality inspection results; Inputting the progress parameter into a preset anomaly detection model to identify and mark abnormal data segments, wherein the anomaly detection model dynamically adjusts the detection threshold based on the characteristics of different anomaly types in the historical engineering data; Input the marked progress parameters into a preset collaborative prediction model to generate a progress trend prediction result, wherein the collaborative prediction model performs multi-dimensional fitting based on task dependencies and resource allocation weights; Using a preset priority scheduling algorithm to classify and integrate the progress trend prediction results and current progress parameters to form a dynamic monitoring data set and store it; The steps of constructing the anomaly detection model include: obtaining a historical engineering data set, wherein each data in the historical engineering data set is marked with an anomaly type and an impact level; dividing training subsets based on the anomaly type and impact level, each training subset corresponding to an anomaly scenario; using the training subsets to train an initial detection model in parallel, until the recognition accuracy of the initial detection model for each anomaly scenario is greater than or equal to a preset first threshold, stopping the training to obtain an intermediate detection model; inputting the historical engineering data set into the intermediate detection model, and verifying whether the anomaly recognition result output by the intermediate detection model meets the preset error range; if so, determining the intermediate detection model as the anomaly detection model.

2. The method for intelligent monitoring of engineering project progress according to claim 1, characterized in that: The real-time acquisition of project progress parameters through the multi-source data acquisition terminal includes: Establishing a communication connection with a target monitoring terminal, wherein the target monitoring terminal is deployed at a preset monitoring node of the engineering project; Continuously reading the real-time data of the target monitoring terminal according to a preset collection period, and marking the collection time point based on the time series characteristics of the real-time data; According to the task topology relationship of the engineering project, the real-time data of different monitoring nodes at the same time point are logically aligned to form a set of associated progress parameters.

3. The method for intelligent monitoring of engineering project progress according to claim 1, characterized in that: Inputting the progress parameter into a preset anomaly detection model includes: Extracting a mutation data segment from the progress parameter, wherein the mutation data segment is a data segment in which the parameter change amplitude exceeds a preset mutation threshold within a continuous monitoring period; Generate an anomaly assessment index based on the duration and degree of deviation of the mutation data segment; The corresponding detection algorithm is dynamically selected according to the anomaly assessment index, wherein the isolation forest algorithm is used for short-term high-amplitude mutations and the cluster analysis algorithm is used for long-term low-amplitude mutations.

4. The method for intelligent monitoring of project progress according to claim 3, characterized in that: The method further comprises: After marking the abnormal data segment, performing data consistency check on the progress parameter; If the verification finds that the data conflict rate exceeds a preset second threshold, the collaborative prediction model is triggered to correct the priority of the conflicting data, wherein the high-priority conflicting data is the task data segment that affects the critical path.

5. The method for intelligent monitoring of engineering project progress according to claim 1, characterized in that: The collaborative prediction model includes the following prediction steps: Constructing a dynamic weight model based on the task dependency network of the engineering project, wherein each task node corresponds to a dependency weight coefficient; Calculate the collaborative prediction weight based on the resource consumption difference of adjacent task nodes; Combined with the historical trends of the progress parameters, multi-dimensional prediction and completion of missing task nodes are performed.

6. The method for intelligent monitoring of engineering project progress according to claim 5, characterized in that: The method further comprises: After the prediction is completed, the prediction results are logically verified, where the verification method includes comparing the execution deviation between the prediction data and the actual task nodes; If the deviation exceeds a preset third threshold, the collaborative prediction weight is readjusted and the prediction is iterated until the deviation is less than the third threshold.

7. The method for intelligent monitoring of engineering project progress according to claim 1, characterized in that: Classifying and integrating the prediction results and current progress parameters using a preset priority scheduling algorithm includes: Classify the first-level classification labels according to the task type, wherein the first-level classification labels include critical tasks, non-critical tasks, and risky tasks; Under each level of classification label, the second level classification sub-label is further divided based on the urgency of the task; The classified task data is stored in different partitions of the time series database according to the label level.

8. The method for intelligent monitoring of engineering project progress according to claim 7, characterized in that: The method further comprises: Configure access levels for classification tags based on pre-set task permissions; Upon receiving a data retrieval request, verify whether the permission identifier provided by the requester matches the access level of the target classification label; If there is a match, the data retrieval channel for the corresponding classification label is opened.

9. An intelligent monitoring system for project progress, characterized in that: include: A multi-source data acquisition module is used to obtain the progress parameters of the engineering project in real time through a multi-source data acquisition terminal, wherein the progress parameters include task completion status, resource consumption, construction period deviation rate and quality inspection results; an anomaly detection module, configured to input the progress parameter into a preset anomaly detection model to identify and mark abnormal data segments, wherein the anomaly detection model dynamically adjusts the detection threshold based on the characteristics of different anomaly types in the historical engineering data; A collaborative prediction module is used to input the marked progress parameters into a preset collaborative prediction model to generate a progress trend prediction result, wherein the collaborative prediction model performs multi-dimensional fitting based on task dependencies and resource allocation weights; Priority scheduling, for classifying and integrating the prediction results and current progress parameters using a preset priority scheduling algorithm to form a dynamic monitoring data set and store it; The steps of constructing the anomaly detection model include: obtaining a historical engineering data set, wherein each data in the historical engineering data set is marked with an anomaly type and an impact level; dividing training subsets based on the anomaly type and impact level, each training subset corresponding to an anomaly scenario; using the training subsets to train an initial detection model in parallel, until the recognition accuracy of the initial detection model for each anomaly scenario is greater than or equal to a preset first threshold, stopping the training to obtain an intermediate detection model; inputting the historical engineering data set into the intermediate detection model, and verifying whether the anomaly recognition result output by the intermediate detection model meets the preset error range; if so, determining the intermediate detection model as the anomaly detection model.

Images