Construction progress control method in project supervision

Abstract

The invention discloses a construction progress control method in project supervision, and the method comprises the steps: calculating the current punctual completion rate and process handover delay time of each subpackage provider according to a real-time progress data set and a pre-established project network diagram, recognizing the progress fluctuation modes of different project stages through a time sequence analysis method, and carrying out the construction progress control according to the progress fluctuation modes. Establishing a correlation model of the sub-packaging quotient and the engineering stage; according to the priority scores, taking minimization of overall construction period delay and resource idleness as objective functions, applying an optimization algorithm, dynamically adjusting a construction plan and a resource allocation scheme of each subpackage provider, and generating an optimized project progress plan which comprises specific work arrangement and resource allocation suggestions of each subpackage provider; a successful coordination strategy in a historical project is analyzed, coordination measures are recommended for project management personnel and a set of communication matrix is generated according to characteristics and progress conditions of a current project, and the coordination measures comprise resource allocation suggestions, process optimization schemes and communication strategy templates.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and in particular to a construction progress control method in engineering supervision. Background Art

[0002] In large and complex engineering projects, there is a key technical contradiction when multiple subcontractors are working at the same time: the on-time completion rate and process handover delay time of each subcontractor vary greatly. This difference is not only related to the subcontractor's own factors, but also affected by the overall progress stage of the project. Specifically, subcontractors in the early stage of the project often face more uncertainties, their on-time completion rate is lower, and the process handover delay time is longer. Subcontractors in the later stage of the project, because the previous work is basically ready, have a relatively high on-time completion rate and a shorter process handover delay time. However, this difference will lead to an imbalance in the overall progress of the project, resulting in a backlog of processes or idle resources on certain critical paths, which in turn affects the construction rhythm of subsequent subcontractors. What is more complicated is that there is a technical dependence on the process connection between different professional subcontractors, and the delay of a subcontractor may have a chain reaction on multiple related subcontractors. This multi-level and multi-dimensional interaction makes it difficult for traditional project progress management methods to effectively deal with it. There is an urgent need for an intelligent management system that can dynamically balance the construction progress of each subcontractor, optimize resource allocation, and coordinate process connection. In addition, this complex project environment also brings severe communication challenges. Information transmission between subcontractors, between subcontractors and general contractors, and between subcontractors and owners is often delayed, distorted, or incomplete. These communication barriers not only exacerbate schedule differences and resource allocation conflicts, but may also lead to wrong decisions and unnecessary conflicts. Traditional unified communication methods are difficult to adapt to the needs and characteristics of different subcontractors at different project stages. Therefore, establishing a customized communication strategy that can be dynamically adjusted according to the characteristics of each party and the project stage has become the key to solving this complex problem. Summary of the invention

[0003] The purpose of the present invention is to solve the problems in the prior art and to propose a construction progress control method in engineering supervision.

[0004] In order to achieve the above object, the present invention adopts the following technical solutions:

[0005] A construction progress control method in engineering supervision comprises the following steps:

[0006] S1, using the data acquisition system to obtain real-time progress data from the construction sites of each subcontractor, transmit it to the central data processing platform, clean, deduplicate and format the collected raw data to obtain a structured real-time progress data set;

[0007] S2, based on the real-time progress data set and the pre-established engineering project network diagram, calculate the current on-time completion rate and process handover delay time of each subcontractor, identify the progress fluctuation pattern of different engineering stages through time series analysis method, and establish the association model between subcontractors and engineering stages;

[0008] S3, based on the engineering construction information, analyze the technical dependency relationship between processes, build a process dependency network diagram, identify bottleneck processes and potential risk points on the critical path, calculate the impact of each process on subsequent construction, generate a process impact factor matrix, and predict the on-time completion rate and process handover delay time of each subcontractor in the future based on the current project characteristics and real-time progress data set, and obtain multi-dimensional progress prediction results;

[0009] S4, based on the correlation model between subcontractor performance and project stage, evaluate the performance expectations of each subcontractor at different stages, combine the process influencing factor matrix, analyze the impact of each process on the overall progress, identify the progress deviation and risk points based on the multi-dimensional progress prediction results, establish a comprehensive scoring model, and obtain the priority score of each subcontractor and process;

[0010] S5, based on the priority scores, with minimizing overall construction delays and resource idleness as the objective function, uses optimization algorithms to dynamically adjust the construction plans and resource allocation plans of each subcontractor, and generates an optimized project schedule, including the work arrangements and resource allocation suggestions for each subcontractor;

[0011] S6, analyze successful coordination strategies in historical projects, and recommend coordination measures and generate a communication matrix for project managers based on the characteristics and progress of the current project. The coordination measures include resource allocation suggestions, process optimization plans, and communication strategy templates.

[0012] S7, based on the optimized schedule plan and coordination strategy, automatically generates task decomposition and progress control indicators for each subcontractor, issues the task decomposition, progress control indicators and communication matrix in real time through the project management information system, and continuously monitors the execution status. When the execution status exceeds the set deviation threshold, an early warning is issued through the preset reporting mechanism and problem escalation process.

[0013] Compared with the prior art, the present invention has the following advantages:

[0014] The invention discloses a construction progress control method in engineering supervision, which solves the problems of untimely progress data collection, low progress prediction accuracy, unreasonable resource allocation, and lack of pertinence of coordination strategies in traditional engineering supervision. Through a real-time data acquisition system and a central data processing platform, efficient collection and standardized processing of real-time progress data on the construction site are realized, providing a reliable data basis for subsequent analysis. By using time series analysis and correlation models, the progress fluctuation pattern is accurately identified and the future progress is predicted, effectively improving the accuracy of progress prediction. Combined with the process dependency network and the influencing factor matrix, the impact of each process on the overall progress is comprehensively evaluated, and more accurate risk identification and priority evaluation are achieved. The construction plan and resource allocation are dynamically adjusted through the optimization algorithm, which significantly improves the resource utilization efficiency. At the same time, based on historical project experience and current project characteristics, customized coordination strategies and communication matrices are generated, which greatly improves the pertinence and effectiveness of project management. Finally, through the automated task decomposition, indicator issuance and execution monitoring mechanism, all-round and refined progress control is achieved, effectively reducing the risk of construction period delays and improving the overall management level of engineering projects. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 A flow chart of a construction progress control method in engineering supervision proposed by the present invention;

[0016] Figure 2 A schematic diagram of a construction progress control method in engineering supervision proposed by the present invention;

[0017] Figure 3 This is another schematic diagram of a construction progress control method in engineering supervision proposed by the present invention. Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0019] In the description of the present invention, it is necessary to understand that the terms "upper", "lower", "front", "back", "left", "right", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention.

[0020] like Figure 1-3 The present embodiment provides a construction progress control method in engineering supervision, which may specifically include:

[0021] Step S1, using the data acquisition system to obtain real-time progress data from the construction sites of each subcontractor, including process completion status, resource usage status, and personnel distribution information, and transmit it to the central data processing platform, and clean, deduplicate, and format the collected raw data to obtain a structured real-time progress data set.

[0022] Real-time progress data with the subcontractor's unique identification number is obtained, including process completion status, resource usage status, and personnel distribution information. The real-time progress data is uploaded by the subcontractor's construction site through mobile applications and spreadsheets; the original data is sent to the central data processing platform using an encrypted transmission protocol based on the real-time progress data. After receiving the original data, the central data processing platform performs preliminary screening and classification; the data cleaning process is performed on the preliminarily sorted data, including: using regular expressions to verify the data format, eliminating outliers and invalid data, and using hash tables to detect duplicate data and eliminate duplicate records; the cleaned data is standardized and formatted according to the preset data format template to obtain a structured real-time progress data set.

[0023] Specifically, when collecting real-time progress data from the construction site, subcontractors fill in the daily workload through a dedicated mobile application, such as 10 cubic meters of concrete pouring, 500 kilograms of steel bar tying, and other specific values. At the same time, the spreadsheet upload function supports batch import of resource usage data such as equipment usage time and material consumption. Personnel distribution information is automatically collected through the positioning devices worn by workers, accurate to within a range of 5 meters. The collected data is encrypted using the AES-256-bit encryption algorithm and transmitted to the central data processing platform via the HTTPS protocol. During the data cleaning process, the regular expression ^[0-9]+(.[0-9]{1,2})? $ is used to verify the numerical format and remove illegal characters and data beyond a reasonable range. The MD5 hash algorithm is used for duplicate data detection. The key fields of each record are spliced ​​together to generate a hash value, and duplicate records are quickly identified by comparing the hash values. Standardization processing converts data in different units uniformly, such as converting hours to minutes.

[0024] Step S2, based on the real-time progress data set and the pre-established project network diagram, calculate the current on-time completion rate and process handover delay time of each subcontractor, identify the progress fluctuation pattern of different project stages through time series analysis method, and establish the association model between subcontractors and project stages.

[0025] The process completion time of each subcontractor is obtained from the real-time progress data set. According to the process completion time and the planned completion time specified in the pre-established project network diagram, the current on-time completion rate of each subcontractor is calculated. If the process completion time is later than the planned completion time, the process handover delay time is recorded. For the on-time completion rate and process handover delay time, the moving average method is used for time series analysis, and the time window of the moving average method is 7 days. According to the results of time series analysis, the progress fluctuation patterns of different engineering stages are classified, and the classification includes dividing the change trend of on-time completion rate into three types: rising, falling, and stable. Based on the on-time completion rate of subcontractors, process handover delay time, and progress feature classification of engineering stages, the decision tree algorithm is used to construct the association model between subcontractors and engineering stages. The decision tree algorithm includes feature selection, tree generation and pruning. The feature selection uses information gain as the standard, the tree generation uses the ID3 algorithm, and the pruning uses the pessimistic pruning method. The accuracy of the association model is evaluated by the cross-validation method. The data that accounts for part of the total data volume is selected as the training set, and the rest is used as the test set. Calculate the prediction accuracy of the association model on the test set to facilitate the adjustment of the association model.

[0026] Specifically, if the actual completion time is later than the planned completion time, the process handover delay time is recorded. The specific calculation method is the difference between the actual completion time of the current process and the planned start time of the next process. For the calculated on-time completion rate and process handover delay time data, the moving average method is used for time series analysis. The time window is set to 7 days, and the average on-time completion rate and average process handover delay time of each subcontractor in different time periods are calculated. The decision tree construction process includes feature selection, tree generation and pruning. Information gain is selected as the feature selection criterion. The decision tree is generated using the ID3 algorithm, and the tree is optimized using the pessimistic pruning method. The accuracy of the association model is evaluated by the cross-validation method. 80% of the data is selected as the training set and 20% as the test set. The prediction accuracy of the model on the test set is calculated.

[0027] Furthermore, in actual application, the completion time of the concrete pouring process of subcontractor A extracted from the real-time progress data set is March 15, 2024, while the planned completion time specified in the project network diagram is March 10, 2024, and the on-time completion rate is calculated to be 83.33%. The calculation of the process handover delay time shows that the actual completion time of the current process is 5 days later than the planned start time of the next steel bar binding process. Applying the moving average method to the data of the past 7 days, it is found that the average on-time completion rate of subcontractor A is 85.71%, and the average process handover delay time is 3.5 days. The progress fluctuation pattern analysis shows that the on-time completion rate of subcontractor A is on the rise, gradually increasing from 80% to 90%.

[0028] Step S3, based on the engineering construction information, analyze the technical dependency relationship between processes, build a process dependency network diagram, identify bottleneck processes and potential risk points on the critical path, calculate the impact of each process on subsequent construction, generate a process impact factor matrix, and predict the on-time completion rate and process handover delay time of each subcontractor in the future based on the current project characteristics and real-time progress data set, and obtain multi-dimensional progress prediction results.

[0029] According to the construction information, the technical dependency relationship between processes is obtained, and the process dependency network diagram is constructed by graph theory algorithm to obtain the adjacency matrix of the process dependency network diagram; the connection relationship between processes is represented by the adjacency matrix, and the critical path method is used to identify the critical path, calculate the earliest start time and the latest start time of each process, and judge the process with zero time margin as the bottleneck process. Comprehensively evaluate the time margin and resource demand of the process, identify potential risk points, set risk thresholds, and mark the process as a high-risk process if the evaluation result of the process exceeds the preset risk threshold; topologically sort the process dependency network diagram, determine the execution order of the processes, calculate the degree of influence of each process on subsequent construction, and use the direct scoring method to construct the influence factor matrix, score according to the duration, resource consumption and technical complexity of the process, and obtain the final influence factor through normalization. The historical on-time completion rate and process handover delay time of each subcontractor are obtained from the real-time progress data set. Combined with the current project characteristics, the exponential smoothing method is used to predict the on-time completion rate and process handover delay time in the future period. The smoothing coefficient α=0.3 is selected, and the prediction formula Ft+1=αXt+(1-α)Ft is used to calculate the future value. The process influencing factor matrix is ​​combined with the predicted on-time completion rate and process handover delay time to build a multi-dimensional progress prediction model. The Monte Carlo simulation method is used to generate a variety of possible progress scenarios. The duration and resource consumption are selected as random variables, and their probability distribution is determined. The number of simulations is set to 1000 times. The total duration and critical path changes of the project under each scenario are calculated to obtain multi-dimensional progress prediction results. The output statistical results include the average duration, the shortest duration, the longest duration and the duration distribution probability.

[0030] Specifically, in practical applications, 20 key processes were extracted from the engineering construction information, and a 20x20 adjacency matrix was constructed to represent the dependencies between processes. The critical path method was used to calculate that the critical path contained 8 processes with a total construction period of 120 days. Three bottleneck processes with zero time margin were identified, namely concrete pouring, steel structure installation, and exterior wall decoration. By evaluating the process time margin and resource requirements, the risk threshold was set to 0.8, and two high-risk processes were identified: scaffolding and curtain wall installation. The direct scoring method was used to construct an impact factor matrix, and the 20 processes were scored with a score range of 1-10. After normalization, the concrete pouring process had the highest impact factor of 0.15. The data of the past 30 days were extracted from the real-time progress data set, and the exponential smoothing method was used for prediction, with the smoothing coefficient α set to 0.3. The prediction results show that the on-time completion rate of the main subcontractor A is expected to be 92%, and the process handover delay time is expected to be 1.5 days. In the multi-dimensional schedule forecasting model, the construction period is set to triangular distribution, with the most likely value being 120 days, the minimum value being 110 days, and the maximum value being 135 days. Resource consumption is set to normal distribution, with the mean being the planned value and the standard deviation being 10% of the planned value. 1,000 Monte Carlo simulations were performed, and the results showed that the average construction period was 122 days, the shortest construction period was 113 days, and the longest construction period was 133 days. Construction period distribution probability analysis shows that there is an 80% probability that the project will be completed within 118-126 days.

[0031] Step S4, based on the correlation model between subcontractor performance and project stage, evaluate the performance expectations of each subcontractor at different stages, combine with the process influencing factor matrix, analyze the impact of each process on the overall progress, identify the progress deviations and risk points based on the multi-dimensional progress prediction results, establish a comprehensive scoring model, and obtain the priority scores of each subcontractor and process.

[0032] According to the correlation model between subcontractor performance and engineering stage, the historical performance data of each subcontractor in different engineering stages are extracted, and the expected performance score of each subcontractor in the future engineering stage is calculated by weighted average method. The weight is set as 0.5 for the data of the most recent month, 0.3 for the first two months, and 0.2 for earlier data, and the expected performance matrix of subcontractors is obtained. Combined with the process influencing factor matrix, the influencing factor of each process is normalized, and the influencing factor is mapped to the score range of 0-100 using logarithmic function mapping, and the impact score of the process on the overall progress is calculated. The difference between the predicted construction period and the planned construction period is extracted from the multi-dimensional progress prediction results, and the progress deviation threshold is set to 5% of the planned construction period. If the difference exceeds the preset progress deviation threshold, it is marked as a progress deviation point. At the same time, combined with the importance of the process and the probability of delay, the risk matrix method is used to identify risk points, and a list of progress deviations and risk points is obtained. A comprehensive scoring model was constructed, and the expected score of subcontractor performance, process impact score, progress deviation and risk point assessment results were used as input variables. The multi-layer perceptron neural network algorithm was used to train the model. The input layer was set with 10 nodes, the hidden layer was set with 20 nodes, and the output layer was set with 1 node. The ReLU function was used as the activation function, the mean square error was used as the loss function, and the Adam optimizer was used as the optimization algorithm. The priority scores of each subcontractor and process were output. The score range was 0-100, and the higher the score, the higher the priority.

[0033] Specifically, in practical applications, the performance data of five major subcontractors in four engineering stages in the past three months are extracted from the association model. For example, the performance score of subcontractor A in the foundation construction stage in the last month is 85 points. The weighted average method is used to calculate the expected future performance, and the expected score of subcontractor A in the next main structure stage is 88 points. The influencing factors of 20 key processes are processed. For example, the original influencing factor of the concrete pouring process is 0.85, and the score is 92 points after being mapped to the 0-100 interval by the logarithmic function. The progress forecast results show that the estimated completion time of the project is 8 days later than planned, exceeding the 5% deviation threshold, and it is marked as a progress deviation point. A 5x5 risk matrix is ​​used to assess process risks, with the horizontal axis being importance (1-5 points) and the vertical axis being delay probability (1-5 points). Processes with scores greater than 20 are marked as high-risk points, and three high-risk processes are identified. The input layer of the comprehensive scoring model contains 5 subcontractor performance expectation scores, 20 process impact scores, 1 progress deviation mark and 3 risk point marks, a total of 29 input nodes. After processing 20 hidden nodes, the final output layer gives 29 priority scores. The model training uses historical project data, a total of 500 samples, 80% of which are used as training sets and 20% as validation sets. After 1,000 rounds of iterative training, the mean square error of the model on the validation set dropped below 0.05. The trained model is used to score the current project, and the priority score of subcontractor A is 92 points, and the priority score of the concrete pouring process is 88 points, which provides the project management team with accurate resource allocation and risk control basis.

[0034] Step S5, based on the priority scores, with minimization of overall construction delays and resource idleness as the objective function, the optimization algorithm is used to dynamically adjust the construction plans and resource allocation plans of each subcontractor, and generate an optimized project schedule, including the specific work arrangements and resource allocation suggestions for each subcontractor.

[0035] According to the priority scores of each subcontractor and process, a multi-objective optimization model is constructed, and the objective function is set to minimize construction delays and idle resources. The construction plan and resource allocation of each subcontractor are used as decision variables. The priority score is incorporated into the objective function by setting the weight coefficient, and a higher priority process is given a greater weight. The construction sequence constraint is expressed as the front-to-back dependency between processes, and the resource constraints include the daily available manpower, equipment quantity, and material supply. The genetic algorithm is used to solve the multi-objective optimization problem. The encoding method uses real number encoding, and each chromosome represents a complete construction plan and resource allocation plan. The population size is set to 100, the crossover probability is 0.8, the mutation probability is 0.1, and the number of iterations is 1000. The fitness function is defined as the inverse of the weighted objective function value, and the weight is determined according to the importance of construction delays and idle resources. The population evolution is carried out through roulette selection, two-point crossover, and uniform mutation operations. The optimal solution obtained by the optimization algorithm is decoded, and the real values ​​in the chromosome are mapped to specific time points and resource quantities to obtain the construction plan and resource allocation plan of each subcontractor. Generate a project schedule in the form of a Gantt chart, including the start time, end time, required resource type and quantity of each process. Set up a regular re-optimization mechanism once a week, and define trigger conditions such as starting dynamic adjustments when the actual progress deviation exceeds 5%. According to the optimized schedule, calculate the workload and resource utilization of each subcontractor, and generate a specific work schedule and resource allocation suggestions. Establish an evaluation and feedback mechanism, collect actual execution data daily, calculate the deviation between the plan and the actual, and trigger the fine-tuning process if the deviation exceeds the preset threshold. The fine-tuning process includes re-executing the optimization process of the multi-objective optimization model to obtain an updated construction plan and resource allocation plan. Set buffer time and backup resources for uncertain factors such as weather impacts and material supply delays, monitor changes in external factors in real time, and adjust optimization parameters in a timely manner.

[0036] Specifically, in actual application, the optimization model is built on the basis of 20 key processes of 5 major subcontractors. The weight of the concrete pouring process with the highest priority score is set to 0.2, and the weights of other processes are allocated according to the priority ratio. The construction sequence constraint is reflected in the fact that the foundation construction must be completed before the main structure, and the resource constraint is set as 200 workers available per day, 10 large equipment, and 500 cubic meters of concrete supply. The chromosome length of the genetic algorithm is 100, the first 80 bits represent the start and end time of the 20 processes, and the last 20 bits represent the resource allocation ratio. The weight of construction delay in the fitness function is 0.6, and the weight of resource idleness is 0.4. After 1000 iterations, the optimal solution shows that the total construction period of the project is 180 days, which is 5 days shorter than the original plan. After decoding, it is obtained that the concrete pouring process is scheduled to start on the 30th day and last for 7 days, requiring 50 workers and 3 pump trucks every day. The generated Gantt chart clearly shows the time schedule and resource requirements of each process. The optimization was set to be automatically triggered at 2 a.m. every Monday. At the same time, the monitoring system detected that the actual progress of the foundation construction was 8% slower than planned, and the dynamic adjustment process was immediately started. The optimized work schedule shows that the workload of subcontractor A in the first week is 1,000 square meters of wall construction, and the required resources include 30 workers and 2 mixers. The evaluation feedback mechanism found that the actual wall construction completed in the first week was 920 square meters, with a deviation of 8%, triggering the fine-tuning process to adjust the workload of the next week to 1,050 square meters. According to the weather forecast, a half-day rainy day buffer time was reserved on the 45th day, and 20% of spare transport vehicles were added during the peak period of material supply on the 60th day to ensure the robustness of the optimization plan.

[0037] Step S6, analyze the successful coordination strategies in historical projects, and recommend coordination measures and generate a communication matrix for project managers based on the characteristics and progress of the current project. The coordination measures include resource allocation suggestions, process optimization plans, and communication strategy templates.

[0038] Obtain the coordination strategy data of successful projects in the historical project database. The coordination strategy data includes resource allocation schemes, process optimization records, and communication strategy implementation. Use the TF-IDF algorithm to extract text features from the coordination strategy data to obtain the coordination strategy feature vector. Based on the coordination strategy feature vector, build a coordination strategy knowledge base. Design the current project feature vector based on the characteristics and progress of the current project. The feature vector includes indicators such as project scale, complexity, progress deviation rate, and resource utilization rate. Combine cosine similarity and Euclidean distance to build a comprehensive similarity index, calculate the similarity between the current project feature vector and the historical project feature vector, and select the historical project with the highest similarity as a reference sample. For example, take the top 10 projects with the highest similarity as reference samples. Perform coordination strategy clustering analysis on the reference samples. Use a hierarchical clustering algorithm to divide the coordination strategies into three categories: resource allocation, process optimization, and communication strategies. Calculate the frequency and success rate of use of each type of strategy and generate a list of recommended coordination measures. Based on the recommended list of coordination measures, use the decision tree algorithm to generate a process optimization plan. Take the process characteristics as input variables and the optimization strategy as output variables to build decision rules. The decision rules are used to guide the process optimization of the current project. Based on the recommended coordination measures, a project communication matrix is ​​constructed. The rows and columns of the communication matrix represent the project stakeholders, and the cell contents include communication frequency, method and key information. The hierarchical analysis method is used to determine the communication importance weight of each stakeholder. A judgment matrix is ​​established to calculate the weight vector. The weight vector is checked for consistency and a communication strategy template is generated. The cross-validation method is used to verify the accuracy of the recommended results. The historical data set is divided into a training set and a test set to evaluate the effectiveness of the recommended coordination measures.

[0039] Specifically, in practical applications, the coordination strategy data of 500 successful projects were extracted from the historical database, and the TF-IDF algorithm was used to process the text to obtain 10,000 feature words, and then these strategies were summarized into 20 topics through the topic model. The current project feature vector contains 10 indicators, such as the project scale is 100,000 square meters, the complexity score is 8.5 points (out of 10 points), the progress deviation rate is -5%, and the resource utilization rate is 85%. In the comprehensive similarity calculation, the cosine similarity weight is 0.6, the Euclidean distance weight is 0.4, and the 10 reference projects with the highest similarity are selected. The hierarchical clustering algorithm sets the clustering threshold to 0.7 and obtains 3 strategy categories: resource allocation (40%), process optimization (35%) and communication strategy (25%). The decision tree algorithm uses the CART method, the tree depth is set to 4, the minimum number of leaf node samples is 5, and 15 process optimization rules are generated. The communication matrix involves 8 stakeholders. The weight vector obtained by the hierarchical analysis method is [0.25, 0.20, 0.15, 0.12, 0.10, 0.08, 0.06, 0.04]. The consistency ratio CR = 0.03 < 0.1, passing the consistency test. The cross-validation adopts the 5-fold method, and the average accuracy rate reaches 85%. Finally, 3 resource allocation suggestions, 5 process optimization rules and 2 communication strategy templates are recommended.

[0040] Step S6 also includes generating a customized communication strategy template for each subcontractor based on the analysis of historical projects and the characteristics of the current project, including recommendations on communication frequency, methods and content, as well as specific reporting mechanisms and problem escalation processes.

[0041] The communication records of subcontractors are extracted from the historical project database, and the communication records are classified using the naive Bayes algorithm to obtain the communication frequency statistics of categories such as progress reports, quality issues, and resource requirements. According to the communication frequency statistics, the communication effect is evaluated based on indicators such as problem resolution time and satisfaction score, and the basic model of communication strategy is constructed. Combined with the characteristics of project scale, complexity, and duration requirements, the content-based recommendation algorithm is used to calculate the similarity between the current project and the historical project, and the historical project with the highest similarity is selected as the reference sample. According to the characteristics of each subcontractor, such as professional field, historical performance, etc., combined with the project stage and work content, the random forest algorithm is used to generate a customized communication strategy, including the recommended communication frequency, preferred communication method, and key communication content. The optimal strategy is determined by the voting results of multiple decision trees. Based on the generated communication strategy, a reporting mechanism and problem escalation process are constructed, and key indicator thresholds are set. When the indicator exceeds the threshold, the escalation process is triggered. The rule engine is used to automatically generate the corresponding report template and escalation path. The rule engine includes rule definition, priority setting, and trigger mechanism. Integrate the generated communication strategy with the project management software to achieve automated reminders and execution, regularly collect feedback data on the effectiveness of strategy execution, and continuously optimize the communication strategy template through machine learning algorithms to improve the applicability and effectiveness of the strategy.

[0042] Specifically, in actual application, 50,000 subcontractor communication records of 1,000 projects were extracted from the historical database, and the naive Bayes algorithm was used to divide these records into five categories: progress report (40%), quality issues (25%), resource requirements (20%), safety issues (10%) and others (5%). The communication effect evaluation is based on the problem solving time (weight 0.6) and satisfaction score (weight 0.4) to obtain a comprehensive score. The current project feature vector contains 8 indicators, such as the project scale of 100,000 square meters, complexity of 8.5 points (out of 10 points), and construction period of 360 days. The content-based recommendation algorithm calculates the similarity with historical projects and selects the 10 most similar projects as references. The random forest algorithm uses 100 decision trees, each with a maximum depth of 5 and a minimum number of leaf node samples of 10, to generate a customized communication strategy. For example, for earthwork subcontractors in the foundation construction phase, it is recommended to make a progress report once a day, use mobile apps as the preferred communication method, and the key communication content includes the volume completed on the day and the plan for the next day. The rule engine sets 20 basic rules, such as triggering the problem escalation process when the daily progress deviation exceeds 10%. After integration with the project management software, communication reminders are automatically sent to subcontractors at 8 am every day. Feedback data shows that after one month of implementation, the average problem resolution time was shortened from 2 days to 1.5 days, and the satisfaction score increased from 7.5 points to 8.2 points (out of 10 points). Based on these feedbacks, the machine learning algorithm automatically fine-tunes the communication strategy every weekend, slightly increasing the communication frequency of high-frequency problems.

[0043] Step S6 also includes extracting the complexity and risk assessment results of the project according to the characteristics and progress of the current project, and generating a dynamically adjusted communication matrix, including stakeholder analysis, communication responsibilities and communication methods.

[0044] Extract the characteristics and progress data of the current project from the project management database, including indicators such as project scale, duration, and completion percentage. Use principal component analysis to calculate the project complexity score, and use PERT (project review and evaluation technique) to quantitatively evaluate project risks and calculate the risk value. Use the complexity score and risk value as input parameters, use the hierarchical analysis method to build a stakeholder influence matrix, and calculate the weight coefficient of each stakeholder. According to the weight coefficient, the stakeholders are divided into three groups: high, medium, and low, and the communication priority and frequency of each group are determined. If the weight coefficient is greater than the preset threshold, the stakeholders are divided into a high priority group, otherwise they are divided into medium and low priority groups. According to the project organizational structure and personnel responsibilities, establish a responsibility allocation matrix, use the adjacency matrix to represent the responsibility relationship network, calculate the node degree to determine the key communication responsible person, and assign the corresponding communication responsible person to each stakeholder group. Combined with the project stage and communication content, use the decision tree algorithm to generate communication method decision rules, automatically select the most suitable communication method according to the real-time project status, and build a dynamically adjusted communication matrix, which includes the corresponding relationship between stakeholders, communication responsible persons, and communication methods. Set a regular update trigger once a week, and trigger the matrix adjustment when key milestone events of the project occur. Establish a feedback mechanism, collect communication effectiveness data through questionnaires, and use regression analysis to continuously optimize the weight coefficients and decision rules in the communication matrix.

[0045] Specifically, in practical applications, 10 key features of the current project were extracted from the project management database, including a project scale of 100,000 square meters, a construction period of 360 days, and a completion percentage of 35%. The principal component analysis method calculated the project complexity score to be 8.5 (out of 10 points). The PERT technique evaluated 50 potential risk points and calculated the project risk value to be 0.65. Based on these two parameters, the hierarchical analysis method constructed a 10x10 stakeholder influence matrix, and calculated 5 high-priority stakeholders (weight>0.15), 3 medium-priority stakeholders (0.05<weight≤0.15), and 2 low-priority stakeholders (weight≤0.05). The responsibility allocation matrix contains 30 roles, and the responsibility relationship network is represented by a 20x20 adjacency matrix. After calculating the node degree, 3 key communication responsible persons are identified. The decision tree algorithm generates 15 communication method decision rules, such as "If the risk value>0.6 and the stakeholder priority is high, daily video conferencing is adopted." The dynamic communication matrix initially contains 50 communication items, which are automatically triggered to update at 2 a.m. every Monday. Additional adjustments are triggered when the project is 30%, 60%, and 90% complete. The feedback mechanism sends an online questionnaire once a month to collect communication satisfaction scores (1-10 points). Regression analysis shows that increasing the communication frequency of high-priority stakeholders from 2 times a week to 3 times a week increases satisfaction by an average of 1.2 points. Based on this, the 4 weight coefficients and 2 decision rules in the communication matrix are automatically adjusted.

[0046] Step S7, based on the optimized schedule plan and coordination strategy, automatically generate task decomposition and progress control indicators for each subcontractor, issue the task decomposition, progress control indicators and communication matrix in real time through the project management information system, and continuously monitor the execution status. When the execution status exceeds the set deviation threshold, an early warning is issued through the preset reporting mechanism and problem escalation process.

[0047] According to the optimized schedule and coordination strategy, the work breakdown structure method is used to decompose the subcontractor's work content. The task time parameters are calculated according to the critical path method combined with resource balance and risk assessment, including the earliest start time, the latest completion time and the total time difference of each task, to obtain the task decomposition tree and progress control indicators. The task decomposition tree, progress control indicators and communication matrix are integrated into the project management information system database and pushed to the subcontractor's mobile terminal through the data interface. RabbitMQ message queue technology is used to ensure the reliability and real-time performance of data transmission. IoT sensors and mobile applications are used to obtain real-time progress data of the subcontractor's construction site, including process completion status, resource usage status and personnel distribution information. The progress data is cleaned and standardized and then stored in the project management information system. The data cleaning process includes outlier detection, missing value processing and format unification. The standardization process uses the Z-score method. Multi-level deviation thresholds are set, and a real-time comparison algorithm is used to monitor the execution status and plan differences. The algorithm calculates the deviation value through data matching and time series analysis. If the deviation exceeds the preset threshold, the report generation program and problem escalation process are triggered, and warning information is issued to the relevant responsible persons through the preset notification channel. Receive reports and problem escalation information, intelligently classify and prioritize warning information, use the naive Bayes algorithm to classify warning content, and determine priority based on urgency and impact scope.

[0048] Specifically, in practical application, a large construction project was task-decomposed and a work decomposition structure tree containing 500 task nodes was generated. The critical path method calculated that the critical path of the project contained 35 tasks with a total construction period of 360 days. After resource balancing, 20% of non-critical tasks were adjusted in time, and risk assessment identified 10 high-risk task points. The task decomposition tree and progress indicators were pushed to the mobile terminals of 50 subcontractors through the RabbitMQ message queue at a rate of 100 messages per second. 200 IoT sensors were deployed on site with a collection frequency of once every 5 minutes, and the mobile application collected 300 progress reports every day. During the data cleaning process, 5% of outliers and 3% of missing values ​​were detected, and 95% of the data fell within ±3 standard deviations after data standardization. The real-time comparison algorithm performed a comprehensive comparison every hour with an average processing time of 30 seconds. Three levels of deviation thresholds were set: 5%, 10% and 20%, and a total of 200 warnings were triggered that month, of which 80% were first-level warnings, 15% were second-level warnings, and 5% were third-level warnings. The Naive Bayes algorithm classified the warnings into three categories: schedule delays, quality issues, and safety hazards, with an accuracy rate of 85%. Priority sorting marked 30% of the warnings as high priority and needed to be handled immediately.

[0049] The above description is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A construction progress control method in engineering supervision, characterized in that: The following steps are involved: S1, using the data acquisition system to obtain real-time progress data from the construction sites of each subcontractor, transmit it to the central data processing platform, clean, deduplicate and format the collected raw data to obtain a structured real-time progress data set; S2, based on the real-time progress data set and the pre-established engineering project network diagram, calculate the current on-time completion rate and process handover delay time of each subcontractor, identify the progress fluctuation pattern of different engineering stages through time series analysis method, and establish the association model between subcontractors and engineering stages; S3, based on the engineering construction information, analyzes the technical dependency relationship between processes, constructs a process dependency network diagram, identifies bottleneck processes and potential risk points on the critical path, calculates the impact of each process on subsequent construction, generates a process impact factor matrix, and combines the current project characteristics and real-time progress data set to predict the on-time completion rate and process handover delay time of each subcontractor in the future, and obtains multi-dimensional progress prediction results; S4, based on the association model between subcontractors and project stages, evaluate the performance expectations of each subcontractor at different stages, combine the process influencing factor matrix, analyze the impact of each process on the overall progress, combine the multi-dimensional progress prediction results, identify the progress deviations and risk points, establish a comprehensive scoring model, and obtain the priority scores of each subcontractor and process; S5, based on the priority scores, with minimization of overall construction delays and resource idleness as the objective function, uses optimization algorithms to dynamically adjust the construction plans and resource allocation plans of each subcontractor to generate an optimized project schedule; S6, analyze successful coordination strategies in historical projects, and based on the characteristics and progress of the current project, recommend coordination measures and generate a communication matrix for project managers; S7, based on the optimized schedule plan and coordination strategy, automatically generates task decomposition and progress control indicators for each subcontractor, issues the task decomposition, progress control indicators and communication matrix in real time through the project management information system, and continuously monitors the execution status. When the execution status exceeds the set deviation threshold, an early warning is issued through the preset reporting mechanism and problem escalation process.

2. A construction progress control method in engineering supervision according to claim 1, characterized in that: The S1 includes: Obtain real-time progress data with subcontractor’s unique identification number, including process completion status, resource usage and personnel distribution information, which is uploaded by subcontractor construction site through mobile applications and electronic forms; According to the real-time progress data, the original data is sent to the central data processing platform using an encrypted transmission protocol. After receiving the original data, the central data processing platform will conduct preliminary screening and classification. Perform data cleaning on the initially collated data, including: using regular expressions to verify the data format and remove outliers and invalid data; Afterwards, a hash table is used to detect duplicate data and eliminate duplicate records; The cleaned data is standardized and formatted according to the preset data format template to obtain a structured real-time progress data set.

3. The construction progress control method in engineering supervision according to claim 2 is characterized in that: The S2 includes: Obtain the process completion time of each subcontractor from the real-time progress data set, and calculate the current on-time completion rate of each subcontractor based on the process completion time and the planned completion time specified in the pre-established engineering project network diagram; If the process completion time is later than the planned completion time, the process handover delay time is recorded; For the on-time completion rate and process handover delay time, the moving average method is used for time series analysis, and the time window of the moving average method is 7 days; Based on the results of time series analysis, the progress fluctuation patterns of different project stages are classified, including the change trend of on-time completion rate is divided into three types: rising, falling and stable; Based on the on-time completion rate of subcontractors, the delay time of process handover and the classification of progress characteristics of engineering stages, the association model between subcontractors and engineering stages was constructed using the decision tree algorithm.

4. The construction progress control method in engineering supervision according to claim 1 is characterized in that: The S3 includes: The technical dependency relationship between processes is obtained based on the engineering construction information, and the process dependency network diagram is constructed using graph theory algorithms to obtain the adjacency matrix of the process dependency network diagram; The connection relationship between processes is represented by the adjacency matrix, the critical path method is used to identify the critical path, and the earliest and latest start times of each process are calculated; Comprehensively evaluate the time margin and resource requirements of the process. If the evaluation result of the process exceeds the preset risk threshold, the process will be marked as a high-risk process. The historical on-time completion rate and process handover delay time of each subcontractor are obtained from the real-time progress data set, and the on-time completion rate and process handover delay time in the future are predicted using the exponential smoothing method; The process influencing factor matrix is ​​combined with the predicted on-time completion rate and process handover delay time to build a multi-dimensional progress prediction model. The Monte Carlo simulation method is used to generate a variety of possible progress scenarios. The total project duration and critical path changes under each scenario are calculated to obtain multi-dimensional progress prediction results.

5. The construction progress control method in engineering supervision according to claim 1 is characterized in that: The S4 includes: Obtain the historical performance data of subcontractors at different engineering stages, and calculate the expected performance score of each subcontractor in the future engineering stage using the weighted average method based on the historical performance data to obtain the subcontractor performance expectation matrix; Normalize the influencing factors of each process in the process influencing factor matrix, map the influencing factors to the score range using logarithmic function mapping, and calculate the impact score of the process on the overall progress; The difference between the predicted duration and the planned duration is extracted from the multi-dimensional progress forecast results. If the difference exceeds the preset progress deviation threshold, it is marked as a progress deviation point. At the same time, the risk matrix method is used to identify risk points based on the importance of the process and the probability of delay, and a list of progress deviations and risk points is obtained. Construct a comprehensive scoring model, taking the subcontractor performance expectation score, process impact score, schedule deviation and risk point assessment results as input variables; The multilayer perceptron neural network algorithm is used to train the comprehensive scoring model. The multilayer perceptron neural network includes an input layer, a hidden layer and an output layer. The input layer receives input variables, and the output layer outputs the priority scores of each subcontractor and process.

6. A construction progress control method in engineering supervision according to claim 5, characterized in that: The S5 includes: A multi-objective optimization model is constructed based on the priority scores of each subcontractor and process. The multi-objective optimization model sets minimization of construction delays and resource idleness as the objective function; Genetic algorithm is used to solve multi-objective optimization problems. The encoding method of genetic algorithm adopts real number encoding. Each chromosome represents a complete construction plan and resource allocation plan. Decode the optimal solution obtained by the genetic algorithm, map the real values ​​in the chromosome to the time points and resource quantities, and obtain the construction plan and resource allocation plan for each subcontractor; Generate a project schedule in the form of a Gantt chart based on the construction plan and resource allocation plan. The project schedule includes the start time, end time, required resource type and quantity of each process.

7. The construction progress control method in engineering supervision according to claim 1 is characterized in that: The S6 includes: Obtain successful project coordination strategy data from the historical project database, including resource allocation plans, process optimization records, and communication strategy implementation; The TF-IDF algorithm is used to extract text features from the coordination strategy data to obtain the coordination strategy feature vector; According to the coordination strategy feature vector, a coordination strategy knowledge base is constructed; Design the current project feature vector, which includes project scale, complexity, schedule deviation rate and resource utilization indicators; The cosine similarity and Euclidean distance are used to construct a comprehensive similarity index to calculate the similarity between the feature vector of the current project and the feature vector of the historical project; Select the historical projects with the highest similarity as reference samples; Conduct coordination strategy cluster analysis on reference samples; A hierarchical clustering algorithm is used to classify coordination strategies into three categories: resource allocation, process optimization, and communication strategies. Calculate the frequency and success rate of each type of strategy and generate a recommended list of coordination measures; Based on the recommended list of coordination measures, a decision tree algorithm is used to generate a process optimization plan; Taking process characteristics as input variables and optimization strategies as output variables, decision rules are constructed; Decision rules are used to guide the process optimization of the current project; Construct a project communication matrix. The rows and columns of the communication matrix represent the project stakeholders, and the cell content includes the communication frequency, method and key information. The analytic hierarchy process is used to determine the communication importance weight of each stakeholder; Establish a judgment matrix and calculate the weight vector; Perform consistency check on weight vectors and generate communication strategy templates; It also includes: generating customized communication strategy templates for each subcontractor based on the analysis of historical projects and the characteristics of the current project, including recommendations on communication frequency, methods and content, as well as specific reporting mechanisms and problem escalation processes; Based on the characteristics and progress of the current project, the complexity and risk assessment results of the project are extracted to generate a dynamically adjusted communication matrix, including stakeholder analysis, communication responsibilities and communication methods.

8. The construction progress control method in engineering supervision according to claim 7 is characterized in that: Based on the analysis of historical projects and the characteristics of the current project, a customized communication strategy template is generated for each subcontractor, including recommendations on communication frequency, methods and content, as well as specific reporting mechanisms and problem escalation processes, including: Obtain subcontractor communication records, classify them using the Naive Bayes algorithm, and obtain communication frequency statistics for progress reports, quality issues, and resource requirements; Based on the statistical results of communication frequency, combined with the project size, complexity, and duration requirements, a content-based recommendation algorithm is used to calculate the similarity between the current project and historical projects, and the historical project with the highest similarity is selected as a reference sample; For the reference sample, a random forest algorithm is used to generate customized subcontractor communication strategies, including recommended communication frequency, preferred communication methods, and key communication content; Based on the communication strategy, build a reporting mechanism and problem escalation process, set key indicator thresholds, and trigger the escalation process if the key indicators exceed the preset thresholds.

9. A construction progress control method in engineering supervision according to claim 8, characterized in that: According to the characteristics and progress of the current project, the complexity and risk assessment results of the project are extracted to generate a dynamically adjusted communication matrix, including stakeholder analysis, communication responsibilities and communication methods, including: The project complexity score is calculated using principal component analysis, and the project risk is quantitatively assessed using PERT to obtain the risk value; According to the project complexity score and risk value, the stakeholder influence matrix is ​​constructed using the hierarchical analysis method to determine the weight coefficient of each stakeholder; If the weight coefficient is greater than the preset threshold, the relevant parties are divided into a high priority group, otherwise they are divided into a medium or low priority group; According to the project organizational structure and personnel responsibilities, a responsibility allocation matrix is ​​established, the adjacency matrix is ​​used to represent the responsibility relationship network, and the node degree is calculated to determine the key communication responsible persons; Use decision tree algorithm to generate communication decision rules, automatically select communication methods according to real-time project status, and build a dynamically adjusted communication matrix; Obtain the correspondence between stakeholders, communication responsible persons and communication methods from the communication matrix, and trigger matrix adjustments when key milestone events of the project occur.

10. A construction progress control method in engineering supervision according to claim 9, characterized in that: The S7 includes: Use the work breakdown structure method to decompose the subcontractor's work content, calculate the task time parameters based on the critical path method combined with resource balance and risk assessment, and obtain the task decomposition tree and progress control indicators; Integrate the task decomposition tree, progress control indicators and communication matrix into the project management information system database and push them to the subcontractor's mobile terminal through the data interface; Use IoT sensors and mobile applications to obtain real-time progress data from subcontractors’ construction sites, clean and standardize the progress data, and store it in the project management information system; Set multi-level deviation thresholds and use real-time comparison algorithms to monitor the difference between execution and plan. If the deviation exceeds the preset threshold, the report generation procedure and problem escalation process will be triggered; Receive reports and problem escalation information, intelligently classify and prioritize warning information, use the naive Bayes algorithm to classify warning content, and determine priority based on urgency and impact scope.

Images