Big data-based decoration design engineering progress management system

Abstract

The application discloses a decoration design engineering progress management system based on big data, which comprises a standardized data acquisition module, an engineering progress risk prediction module, an engineering task node time optimization module and an engineering progress intelligent management module. The application relates to the technical field of data processing, in particular to a decoration design engineering progress management system based on big data. The application innovatively proposes a task node dynamic time window reconstruction method based on overall delay result reverse mapping, improves the accuracy and automation degree of the conversion of the overall delay prediction result into a specific construction time adjustment scheme, proposes an engineering progress risk prediction model based on the combination of a graph neural network and a bidirectional long short-term memory network, realizes accurate overall delay risk prediction of the decoration design engineering, and adopts a task node time parameter search method based on double-mode speed updating and particle global optimal position hybrid mutation improvement, so as to realize adaptive optimization of the decoration design engineering progress plan.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a decoration design project progress management system based on big data. Background Technology

[0002] The Decoration Design Project Progress Management System is a full-process intelligent project progress control system for decoration companies, design studios, and owners. Relying on big data processing, artificial intelligence, the Internet of Things, and mobile Internet technologies, it uniformly collects, standardizes, and integrates heterogeneous data from multiple sources throughout the entire lifecycle of decoration design projects. This enables intelligent early warning of project progress deviations and intelligent prediction of project duration, effectively solving industry pain points such as information asymmetry, chaotic change control, and frequent project delays in traditional decoration project progress management. It improves the precision and intelligence of decoration project progress control and shortens project delivery cycles.

[0003] However, traditional methods for managing the progress of interior design projects typically only identify or issue early warnings about the overall project delay risk. They struggle to translate the overall delay predictions into time adjustment plans for specific task nodes and often rely on manual extensions or simple compressions of the schedule. This leads to low efficiency and unstable adjustment results in interior design project progress management. Existing models for predicting progress risks in interior design projects also struggle to simultaneously capture the high-order dependencies between task nodes and the dynamic changes in the time series of construction progress, resulting in low accuracy in predicting overall delay risks. Furthermore, existing methods for optimizing project progress often rely on fixed rules and manual experience to adjust schedule nodes, leading to unreasonable adjustment results and a failure to effectively reduce the overall project delay risk. Summary of the Invention

[0004] To address the above issues and overcome the shortcomings of existing technologies, this invention provides a big data-based decoration design project progress management system. Traditional methods for decoration design project progress management typically only identify or issue early warnings of overall project delay risks, failing to translate overall delay predictions into specific task node time adjustment plans. Furthermore, they often rely on manual adjustments such as extending or simply compressing the schedule, leading to low efficiency and unstable adjustment results. This invention innovatively proposes a dynamic time window reconstruction method for task nodes based on a reverse mapping of overall delay results. This improves the accuracy and automation of converting overall delay predictions into specific construction time adjustment plans, enhances the adaptability of the decoration design project schedule, and improves the rationality and feasibility of task node time adjustments. It achieves dynamic optimization management of decoration project task node times driven by overall delay risk prediction. Addressing the problem that existing models for predicting decoration design project progress risks struggle to simultaneously capture high-order dependencies between task nodes and the dynamic changes in the time series of construction progress, resulting in low overall delay risk prediction accuracy, this invention innovatively proposes a method based on graph neural networks and bidirectional length-length... This invention proposes a project schedule risk prediction model combining a period memory network. By constructing a task node graph and extracting high-order interaction features of nodes, and utilizing bidirectional LSTM to extract dynamic features of the construction time series, it improves the accuracy and stability of overall project delay risk prediction, enhances the model's adaptability to dynamic factors such as construction node anomalies, and improves the guidance of the prediction results for the task node time optimization module. This enables accurate prediction of overall delay risk for decoration design projects. Addressing the technical problem that existing project schedule optimization methods often rely on fixed rules and manual experience for adjusting schedule node times, leading to unreasonable adjustment results and ineffective reduction of overall project delay risk, resulting in chaotic and inefficient project schedule management, this invention proposes an improved task node time parameter optimal search method based on a hybrid mutation of dual-mode velocity update and particle global optimal position. This improves the global optimization capability and local adjustment accuracy of task node time parameter search, enhances the adaptability to complex decoration construction constraints and dynamic schedule changes, and improves the rationality, stability, and executability of the updated project schedule. It achieves adaptive optimization of the decoration design project schedule with the goal of minimizing the overall delay days.

[0005] The technical solution adopted by the present invention is as follows: The decoration design project progress management system based on big data provided by the present invention includes a standardized data acquisition module, a project progress risk prediction module, a project task node time optimization module, and a project progress intelligent management module;

[0006] The standardized data acquisition module specifically obtains standardized data for decoration design project progress management through data collection and data preprocessing.

[0007] The project progress risk prediction module is used to predict the overall delay risk of the entire decoration project and provide a basis for decision-making for task node time optimization. Specifically, it first constructs a project progress risk prediction model that can capture the high-order dependencies between task nodes and the dynamic changes of time series, trains the project progress risk prediction model, and finally performs real-time prediction of project progress risk to obtain the overall real-time delay days of the project.

[0008] The project task node time optimization module is used to dynamically adjust the start and end times of each unfinished task node in the project, so as to realize adaptive optimization of the decoration design project schedule. Specifically, it includes filtering unfinished task nodes, obtaining a set of time parameters of nodes to be optimized, performing optimal search for task node time parameters based on an optimization algorithm improved by dual-mode velocity update and particle global optimal position hybrid mutation, reconstructing task node dynamic time windows and updating the project schedule, thereby obtaining the updated decoration design project schedule.

[0009] The intelligent project progress management module first performs real-time prediction of project progress risks, obtains the total real-time delay days of the current decoration design project, and judges the delay risk based on the total real-time delay days. If there is a delay risk, the updated decoration design project progress plan is obtained through the project task node time optimization module; otherwise, the original progress plan is maintained, thereby realizing intelligent management of the decoration design project progress.

[0010] Furthermore, the standardized data acquisition module specifically collects raw data for decoration design project progress management and performs data preprocessing to obtain standardized data for decoration design project progress management.

[0011] The standardized data for the progress management of decoration design projects includes standardized data for historical project progress risk prediction and standardized data for real-time project progress risk prediction.

[0012] The data preprocessing specifically involves cleaning, normalizing, and selecting data features from the original data of decoration design project progress management to obtain standardized data for decoration design project progress management.

[0013] Furthermore, the project schedule risk prediction module specifically includes the following steps:

[0014] The construction of a project schedule risk prediction model includes the following steps:

[0015] Constructing a task node graph involves treating each engineering task node as a node in the graph, building node features for each node, constructing an edge set and a node-edge adjacency matrix, and obtaining the task node graph through the construction of node features and adjacency matrix.

[0016] The project task node refers to an independent construction task unit formed by decomposing the decoration design project according to the construction process, construction area, type of work and construction sequence.

[0017] The edge connection rules in the edge set include the dependencies between task nodes, conflicts in construction areas, and the connection between work processes.

[0018] The high-order interaction feature extraction specifically involves inputting the task node graph into a graph neural network and using graph convolution operations to aggregate the node features to obtain the high-order aggregated features of each node.

[0019] Temporal augmentation feature extraction specifically involves arranging the high-order aggregated features of each node according to the time series to form an input sequence, which is then input into a bidirectional long short-term memory network for forward and reverse temporal modeling to obtain the temporal augmentation features of each node.

[0020] The project schedule risk prediction output specifically involves average pooling the features of all nodes at the last time step to obtain the global feature vector of the overall project schedule, and then inputting it into the multilayer perceptron regression network to output the overall delay days of the project.

[0021] The training of the project schedule risk prediction model involves using standardized historical project schedule risk prediction data as model training data to iteratively train the prediction model and obtain the trained project schedule risk prediction model.

[0022] Real-time prediction of project schedule risk involves inputting standardized real-time project schedule risk prediction data into a trained project schedule risk prediction model to obtain the overall real-time delay days for the project.

[0023] Furthermore, the project task node time optimization module specifically includes the following steps:

[0024] The incomplete task node filtering is used to identify the construction task nodes that have not yet been completed in the decoration design project, and to provide adjustment objects for task node time optimization; specifically, based on the task node status identification results, the nodes under construction and the nodes not under construction are filtered out from all task nodes to form a set of incomplete task nodes.

[0025] Obtain the set of time parameters for the task nodes to be optimized. Specifically, based on the set of incomplete task nodes, take the start time and end time of each incomplete task node as the parameters to be optimized to obtain the set of time parameters for the task nodes to be optimized.

[0026] The optimal search for task node time parameters specifically employs an improved optimization algorithm based on a hybrid mutation of dual-mode velocity updates and global optimal particle positions to dynamically optimize the start and end times of each incomplete task node, thereby obtaining the optimal set of task node time parameters with the fewest delay days. This includes the following steps:

[0027] Initializing the search population involves encoding the set of time parameters of the task nodes to be optimized into the position vectors of the search individuals in the optimization algorithm, thereby completing the initialization of the particle population.

[0028] The particle fitness value is calculated by replacing the start and end times of the corresponding task nodes in the real-time engineering progress risk prediction standardized data with the set of task node time parameters corresponding to the particle's position vector, obtaining candidate engineering progress risk prediction standardized data, inputting the candidate engineering progress risk prediction standardized data into the trained engineering progress risk prediction model, obtaining the overall project delay days corresponding to the particle, and using the overall project delay days as the fitness evaluation value of the particle.

[0029] The global optimal position of the particle undergoes a hybrid mutation process. First, based on the ratio of the current iteration count to the maximum iteration count, adaptive weighting coefficients for two mutation modes are calculated. Then, Gaussian random numbers following a standard normal distribution and Cauchy random numbers following a standard Cauchy distribution are generated. Finally, the global optimal position of the particle is vector-superimposed with the adaptively weighted mutation perturbation term to obtain the mutated global optimal position of the particle. ;

[0030] Particle updates specifically involve generating an independent uniform random number in the interval [0,1] for each particle. Based on random numbers and preset dwell probability Based on the magnitude relationship, two differentiated velocity updates are selected: if the random number is less than the dwell probability, a dwell behavior velocity update is performed; otherwise, a jump-type particle velocity update is performed. Then, the particle position is updated based on the updated particle velocity. The formula used is as follows:

[0031] ;

[0032] In the formula, Indicates the current iteration number. Indicates the first Inertia weights in the next iteration It indicates that the i-th particle is in the... Speed ​​in the next iteration It indicates that the i-th particle is in the... Speed ​​in the next iteration This represents the local optimal position of an individual particle. , and Represents a random number in the range [0,1]. It indicates that the i-th particle is in the... Position in the next iteration and These represent individual learning factors and group learning factors, respectively. Let d represent the Villevian flight step size vector, where d represents the dimension of the solution space of the optimization problem, corresponding to the number of variables to be optimized. This represents generating a 1-row vector with d columns and all elements being 1;

[0033] The particle search terminates by first updating the optimal particle position, and then terminating the search and obtaining the global optimal particle position when the particle fitness value is less than the fitness threshold or when the maximum number of iterations is reached; the global optimal particle position specifically refers to the optimal set of task node time parameters.

[0034] The optimal position update of the particle specifically involves re-evaluating the fitness value of all updated particles and comparing it with the current global optimal position of the particle based on the current particle's fitness value. If the current particle's fitness value is better, then the global optimal position of the particle is updated.

[0035] The task node dynamic time window reconstruction specifically involves using the start and end times of each task node in the optimal task node time parameter set as the recommended start and end times for each incomplete task node, thereby generating a dynamic time window for each incomplete task node.

[0036] The project schedule is updated by replacing the planned start and end times of the corresponding task nodes in the original project schedule with the recommended start and end times of each unfinished task node based on the dynamic time windows of each unfinished task node, while keeping the actual completion time of completed task nodes and the actual start time of nodes under construction unchanged, thus obtaining the updated decoration design project schedule.

[0037] Furthermore, the intelligent project progress management module first calls the project progress risk prediction module to perform real-time prediction of the project progress risk of the current decoration design project, and obtains the overall real-time delay days of the current decoration design project. Based on the overall real-time delay days, it determines whether the current decoration design project has a delay risk. When it is determined that the current decoration design project has a delay risk, the updated decoration design project progress plan is obtained through the project task node time optimization module. Construction progress management is carried out according to the project progress plan. Otherwise, the current decoration design project progress plan is maintained, thereby realizing intelligent management of the decoration design project progress.

[0038] The beneficial effects achieved by the present invention using the above solution are as follows:

[0039] (1) In view of the technical problems of traditional decoration design project progress management methods, which can only identify or issue early warnings of the overall project delay risk, it is difficult to transform the overall delay prediction results into specific task node time adjustment schemes. Moreover, they often rely on manual extension or simple compression of the construction period for progress adjustment, resulting in low efficiency and unstable adjustment results in decoration design project progress management. This invention innovatively proposes a task node dynamic time window reconstruction method based on the reverse mapping of the overall delay result. After obtaining the overall real-time delay days of the project, the unfinished task nodes are screened through the project task node time optimization module, distinguishing between nodes under construction and unfinished task nodes. Based on the set of time parameters of the task nodes to be optimized, the optimal time parameters are searched to obtain the recommended start time and recommended end time of each unfinished task node. This improves the accuracy and automation of the transformation of the overall delay prediction results into specific construction time adjustment schemes, enhances the adaptability of the decoration design project progress plan, improves the rationality and feasibility of task node time adjustment, and realizes the dynamic optimization management of decoration project task node time driven by the overall delay risk prediction.

[0040] (2) In view of the technical problem that existing models for predicting the progress risk of decoration design projects have difficulty capturing the high-order dependencies between task nodes and the dynamic changes in the time series of construction progress at the same time, resulting in low overall delay risk prediction accuracy, this invention innovatively proposes a project progress risk prediction model based on the combination of graph neural network and bidirectional long short-term memory network. By constructing a task node graph and extracting the high-order interaction features of the nodes, and by using bidirectional LSTM to extract the dynamic features of the construction time series, the accuracy and stability of the overall project delay risk prediction are improved, the model's adaptability to dynamic factors such as construction node anomalies is enhanced, and the guidance role of the prediction results for the task node time optimization module is improved, thus realizing accurate prediction of the overall delay risk of decoration design projects.

[0041] (3) In view of the technical problems that existing methods for optimizing project schedules often use fixed rules and manual experience to adjust the schedule node time, resulting in unreasonable schedule adjustment results and the failure to effectively reduce the risk of overall project delay, which leads to chaotic project schedule management and low efficiency, this invention proposes a task node time parameter optimal search method based on dual-mode speed update and particle global optimal position hybrid mutation. This improves the global optimization capability and local adjustment accuracy of task node time parameter search, enhances the adaptability to complex decoration construction constraints and dynamic schedule changes, improves the rationality, stability and executability of the updated project schedule, and realizes adaptive optimization of the decoration design project schedule with the goal of minimizing the overall delay days. Attached Figure Description

[0042] Figure 1 A schematic diagram of the modules of the decoration design project progress management system based on big data provided by the present invention;

[0043] Figure 2 A flowchart illustrating the process of building a project schedule risk prediction model in the project schedule risk prediction module;

[0044] Figure 3 A flowchart illustrating the module for optimizing project task node times;

[0045] Figure 4 A flowchart illustrating the optimal search process for task node time parameters in the task node time optimization module of the project;

[0046] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. Detailed Implementation

[0047] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0048] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the system or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0049] Example 1, see Figure 1 The big data-based decoration design project progress management system provided by this invention includes a standardized data acquisition module, a project progress risk prediction module, a project task node time optimization module, and a project progress intelligent management module.

[0050] The standardized data acquisition module specifically obtains standardized data for decoration design project progress management through data collection and data preprocessing, and sends the data to the project progress risk prediction module and the project task node time optimization module.

[0051] The project progress risk prediction module receives data from the standardized data acquisition module and the project task node time optimization module to predict the overall delay risk of the entire decoration project and provide a basis for decision-making for task node time optimization. Specifically, it first constructs a project progress risk prediction model that can capture the high-order dependencies and dynamic changes in time series between task nodes, trains the project progress risk prediction model, and finally performs real-time prediction of project progress risk to obtain the overall real-time delay days of the project, and sends the data to the project task node time optimization module and the project progress intelligent management module.

[0052] The project task node time optimization module receives data from the standardized data acquisition module and the project progress risk prediction module. It is used to dynamically adjust the start and end times of each unfinished task node in the project, so as to realize adaptive optimization of the decoration design project schedule. Specifically, it includes screening unfinished task nodes, obtaining the set of time parameters of the nodes to be optimized, performing optimal search for task node time parameters based on the optimization algorithm improved by dual-mode speed update and particle global optimal position hybrid mutation, reconstructing the dynamic time window of the task node, and updating the project schedule, thereby obtaining the updated decoration design project schedule, and sending the data to the project progress intelligent management module and the project progress risk prediction module.

[0053] The intelligent project progress management module receives data from the project progress risk prediction module and the project task node time optimization module. Specifically, it first performs real-time prediction of project progress risk, obtains the overall real-time delay days of the current decoration design project, and judges the delay risk based on the overall real-time delay days. If there is a delay risk, the updated decoration design project progress plan is obtained through the project task node time optimization module; otherwise, the original progress plan is maintained, thereby realizing intelligent management of the decoration design project progress.

[0054] By performing the above operations, this invention addresses the technical problems of traditional decoration design project progress management methods, which typically only identify or issue early warnings of overall project delay risks, making it difficult to translate overall delay predictions into specific task node time adjustment plans. Furthermore, these methods often rely on manual extensions or simple compression of the construction period for schedule adjustments, resulting in low efficiency and unstable adjustment results in decoration design project progress management. This invention innovatively proposes a dynamic time window reconstruction method for task nodes based on the reverse mapping of overall delay results. After obtaining the real-time delay days for the entire project, the method uses a project task node time optimization module to filter incomplete task nodes, distinguishing between nodes under construction and those not yet under construction. Based on the set of time parameters for the task nodes to be optimized, it searches for optimal time parameters to obtain recommended start and end times for each incomplete task node. This improves the accuracy and automation of converting overall delay predictions into specific construction time adjustment plans, enhances the adaptability of decoration design project schedules, and improves the rationality and feasibility of task node time adjustments. It achieves dynamic optimization management of decoration project task node times driven by overall delay risk prediction.

[0055] Example 2, see Figure 1 This embodiment is based on the above embodiment. Specifically, the standardized data acquisition module collects data through the decoration design project management platform to obtain the original data of decoration design project progress management, and performs data preprocessing to obtain standardized data of decoration design project progress management.

[0056] The original data for the decoration design project progress management includes historical project progress risk prediction data and real-time project progress risk prediction data.

[0057] The standardized data for the progress management of decoration design projects includes standardized data for historical project progress risk prediction and standardized data for real-time project progress risk prediction.

[0058] Both the historical project progress risk prediction data and the real-time project progress risk prediction data include basic project attribute data, project planned progress data, project actual progress data, task node status data, material supply status data, design change status data, customer confirmation status data, and on-site construction anomaly data.

[0059] The basic attribute data of the project includes project type, project area, number of construction spaces, decoration style, decoration level, total contract period, contract start time, contract completion time, project budget amount, number of construction teams, and type of construction work.

[0060] The project schedule data includes the project start time, project end time, planned start time of each project task node, planned end time of each task node, planned duration of each task node, dependencies between task nodes, and overall completion rate of the plan.

[0061] The project task nodes refer to independent construction task units formed by decomposing the decoration design project according to the construction process, construction area, type of work and construction sequence. These units are independently manageable, have a definite start and end time, and can track the construction status. They include demolition and alteration project nodes, water and electricity construction nodes, masonry construction nodes, carpentry construction nodes, painting construction nodes, equipment installation nodes, soft furnishing arrival nodes, and completion acceptance nodes.

[0062] The actual progress data of the project includes the current construction date, the current number of days of construction, the current overall completion rate, the current remaining construction period, the current number of completed task nodes, the current number of uncompleted task nodes, the actual start time, the actual completion time, the actual completion rate, and the current construction status of each task node;

[0063] The task node status data includes the task node name, the construction area to which the task node belongs, the type of work corresponding to the task node, whether the task node is a preceding node, whether the task node has the conditions for parallel construction, the remaining construction period of the task node, the compressible construction period of the task node, and the construction sequence constraints of the task node.

[0064] The material supply status data includes material name, material category, corresponding task node, planned arrival time, actual arrival time, estimated arrival time, arrival deviation days, inventory status, production cycle of customized materials, historical on-time rate of material suppliers, and feasibility of material substitution.

[0065] The design change status data includes the number of design changes, the time of the design change, the construction area to which the design change belongs, the work process involved in the design change, the processing time of the design change, the number of task nodes affected by the design change, and the project duration affected by the design change.

[0066] The customer confirmation status data includes the type of customer confirmation item, the time the customer confirmation item was initiated, the time the customer confirmation was completed, the waiting time for the customer confirmation, the number of times the customer confirmation was overdue, the task node corresponding to the customer confirmation item, and the project schedule affected by the customer confirmation item.

[0067] The on-site construction anomaly data includes the type of construction anomaly, the time of occurrence of the construction anomaly, the duration of the construction anomaly, the construction area to which the construction anomaly belongs, the corresponding task node of the construction anomaly, the handling status of the construction anomaly, and the impact of the construction anomaly on the construction period.

[0068] The historical project schedule risk prediction data also includes the results of historical project schedule risks.

[0069] The specific result of the historical project progress risk is the actual number of days of delay for the historical decoration design project. The actual number of days of delay is calculated based on the actual completion time and the contractually agreed completion time of the historical project. When the number of days of delay is positive, it indicates that a delay has occurred; when the number of days of delay is zero, it indicates that the historical project was completed on schedule; when the actual number of days of delay is negative, it indicates that the historical project was completed ahead of schedule.

[0070] The data preprocessing is used to improve the integrity, accuracy, uniformity and usability of the original data for decoration design project progress management. Specifically, it involves data cleaning, normalization and data feature selection of the original data for decoration design project progress management to obtain standardized data for decoration design project progress management.

[0071] The data cleaning process is used to remove erroneous, duplicate, and invalid data from the original data of the decoration design project progress management, thereby improving the reliability of subsequent project progress risk prediction and task node time optimization. Specifically, it involves removing outliers, deleting duplicate data, and processing missing data in the original data of the decoration design project progress management.

[0072] The outlier removal specifically involves identifying data that exceeds the preset reasonable value range or significantly deviates from the statistical distribution characteristics based on the preset reasonable value range of each data field, and removing the identified outlier data.

[0073] The deduplication process specifically involves identifying duplicate project progress data based on the project number, task node number, data collection timestamp, construction area identifier, and data source identifier, and retaining the most recent and most complete valid data from the duplicate data.

[0074] The missing data processing specifically involves identifying missing fields in the original data of the decoration design project progress management and processing them according to the data type of the missing fields; for continuous fields, the mean value is used for completion, and for categorical fields, the mode is used for completion.

[0075] The normalization process is used to convert raw data from different sources, of different types, and with different dimensions into a unified data representation. Specifically, it first uses label coding or one-hot coding to convert categorical fields in the raw data into corresponding numerical features based on their different categories. Then, it uses the Z-Score standardization algorithm to normalize continuous variables, converting each continuous variable into a standardized variable with a mean of 0 and a standard deviation of 1.

[0076] The data feature selection is used to filter out features with high correlation to the project schedule risk prediction results from the normalized decoration design project schedule management data, reduce redundant and noisy features, and improve the training efficiency, prediction accuracy and interpretability of the project schedule risk prediction model. Specifically, it uses correlation analysis to calculate the linear correlation coefficient between each feature and the project schedule risk prediction results, and filters out core features with an absolute value of correlation coefficient greater than a preset threshold.

[0077] Example 3, see Figure 1 and Figure 2 This embodiment is based on the above embodiment. The project progress risk prediction module is used to predict the overall delay risk of the entire decoration project, providing a basis for decision-making for task node time optimization; specifically, it includes the following steps:

[0078] A project schedule risk prediction model is constructed to fully capture the high-order correlations and dynamic changes in the time series among various task nodes in the decoration design project, thereby achieving accurate prediction of the overall project delay risk. The specific steps include:

[0079] A task node graph is constructed to model the process dependencies and sequence relationships between task nodes and capture higher-order node associations. Specifically, each engineering task node is treated as a node in the graph, and node features are constructed for each node. An edge set and a node-edge adjacency matrix are also constructed. The task node graph is obtained through the construction of node features and adjacency matrix.

[0080] The node features include planned construction period, current completion rate, remaining construction period, material arrival status, design change status, construction area information, and customer confirmation status.

[0081] The edge connection rules in the edge set include the dependencies between task nodes, conflicts in construction areas, and the connection between work processes.

[0082] The node-edge adjacency matrix is ​​used to represent the graph structure, wherein the elements of the adjacency matrix are... This indicates whether there is an edge connecting node i and node j. If node i and node j have a dependency relationship, a conflicting construction area, or a connection between work processes, then... =1, otherwise =0;

[0083] High-order interaction feature extraction is used to extract the high-order interaction features of task nodes in the entire engineering graph, enhance the information transmission between nodes, and improve the model's ability to perceive the impact of local delays on overall delays. Specifically, the task node graph is input into a graph neural network, and the node features are aggregated using graph convolution operations to obtain the high-order aggregated features of each node.

[0084] Temporal augmentation feature extraction is used to capture the dynamic changes in the construction progress of each task node over time, thereby enhancing the ability to predict future progress trends. Specifically, the high-order aggregated features of each node are arranged in time series to form an input sequence, which is then input into a bidirectional long short-term memory network for forward and reverse time modeling to obtain the temporal augmentation features of each node.

[0085] The time-series arrangement specifically involves arranging the high-order aggregated features of task nodes corresponding to different collection times of the same decoration design project according to the time sequence based on the data collection timestamps of the project progress data.

[0086] The project schedule risk prediction output specifically involves average pooling the features of all nodes at the last time step to obtain the global feature vector of the overall project schedule, and then inputting it into the multilayer perceptron regression network to output the overall delay days of the project.

[0087] The training of the project schedule risk prediction model involves using standardized historical project schedule risk prediction data as model training data to iteratively train the prediction model and obtain the trained project schedule risk prediction model.

[0088] The iterative training of the prediction model specifically employs the mean squared error loss function, uses historical project progress risk results as supervision labels, and iteratively updates the weight matrix and bias parameters of the prediction model through backpropagation algorithm and gradient descent optimization method. The prediction model parameters are continuously optimized through multiple rounds of iteration. When the preset maximum number of training times is reached or the loss function value converges to a set threshold, the iterative training stops.

[0089] Real-time prediction of project schedule risk involves inputting standardized real-time project schedule risk prediction data into a trained project schedule risk prediction model to obtain the overall real-time delay days for the project.

[0090] By performing the above operations, this invention addresses the technical problem that existing models for predicting the progress risk of decoration design projects struggle to simultaneously capture the high-order dependencies between task nodes and the dynamic changes in the construction progress over time, resulting in low overall prediction accuracy for delay risks. This invention innovatively proposes a project progress risk prediction model based on a combination of graph neural networks and bidirectional long short-term memory networks. By constructing a task node graph and extracting high-order interaction features of the nodes, and by utilizing bidirectional LSTM to extract dynamic features of the construction time series, the accuracy and stability of overall project delay risk prediction are improved. This enhances the model's adaptability to dynamic factors such as construction node anomalies, improves the guidance role of the prediction results for the task node time optimization module, and achieves accurate prediction of overall delay risks for decoration design projects.

[0091] Example 4, see Figure 1 , Figure 3 and Figure 4 This embodiment is based on the above embodiment. The project task node time optimization module is used to dynamically adjust the start and end times of each unfinished task node in the project, thereby achieving adaptive optimization of the decoration design project schedule. Specifically, it includes the following steps:

[0092] The incomplete task node filtering is used to identify the construction task nodes that have not yet been completed in the decoration design project, and to provide adjustment objects for task node time optimization; specifically, based on the task node status identification results, the nodes under construction and the nodes not under construction are filtered out from all task nodes to form a set of incomplete task nodes.

[0093] The task node status identification result is specifically based on the actual completion ratio and actual construction status of each task node to identify completed task nodes, construction nodes, and non-construction task nodes. Among them, task nodes with a completion ratio of 100% and a construction status of completed are not included in the set of non-complete task nodes, ensuring that subsequent time optimization only applies to task nodes that have not yet been completed.

[0094] Obtain a set of time parameters for task nodes to be optimized. This set is used to determine the time parameters that need to be adjusted during the optimization process of task node times, so as to dynamically optimize incomplete task nodes. Specifically, based on the set of incomplete task nodes, the start and end times of each incomplete task node are used as parameters to be optimized, resulting in a set of time parameters for task nodes to be optimized. For task nodes that are currently under construction in the set of incomplete task nodes, their start time is fixed as the actual start time, and only the end time is used as a parameter to be optimized. For task nodes that have not yet started construction, both their start and end times are parameters to be optimized.

[0095] The optimal search for task node time parameters is used to obtain the optimal construction time arrangement for each incomplete task node, thereby minimizing the overall delay days. Specifically, an optimization algorithm based on a hybrid mutation of dual-mode velocity update and particle global optimal position is used to dynamically optimize the start and end times of each incomplete task node, obtaining the optimal set of task node time parameters with the fewest delay days. This includes the following steps:

[0096] Initialize the search population to generate the initial search solution set for the optimization algorithm, providing a basis for candidate solutions for dynamic optimization. Specifically, encode the set of time parameters of the task nodes to be optimized into position vectors of search individuals in the optimization algorithm, and randomly generate the position vectors of M individual particles. This completes the initialization of the particle population;

[0097] Each individual particle represents a candidate task node time parameter, and each dimension in the vector corresponds to the start and end time of an incomplete task node.

[0098] The particle fitness value is calculated to evaluate the effect of each candidate task node time scheme on the overall project delay day optimization target. Specifically, the set of task node time parameters corresponding to the particle's position vector is used to replace the start and end times of the corresponding task nodes in the real-time project progress risk prediction standardized data to obtain candidate project progress risk prediction standardized data. The candidate project progress risk prediction standardized data is then input into the trained project progress risk prediction model to obtain the overall project delay days corresponding to the particle. The overall project delay days are then used as the fitness evaluation value of the particle.

[0099] Hybrid mutation of the global optimal position of particles addresses the core flaw in traditional particle swarm optimization (PSO) algorithms: the concentration of all particles at a single global optimal position, leading to rapid loss of population diversity and susceptibility to local optima. It applies an adaptive hybrid mutation perturbation to the current global optimal position, guiding the population out of local extreme regions. Specifically, it first calculates adaptive weight coefficients for two mutation modes based on the ratio of the current iteration count to the maximum iteration count, generating Gaussian random numbers following a standard normal distribution and Cauchy random numbers following a standard Cauchy distribution, respectively. Then, it vector-superimposes the global optimal position of the particles with the adaptively weighted mutation perturbation term to obtain the mutated global optimal position of the particles. The formula used is as follows:

[0100] ;

[0101] In the formula, Indicates the current iteration number. This represents the global optimal position of the particle. Indicates the maximum number of iterations. Let represent a d-dimensional random vector that follows a standard normal distribution, where each component independently follows a normal distribution with a mean of 0 and a variance of 1. This is used to implement local fine perturbations of Gaussian variation. Let represent a d-dimensional random vector that follows a standard Cauchy distribution, with each component independently following a standard Cauchy distribution, used to realize long-range global perturbations of Cauchy variation;

[0102] Particle updates specifically involve generating an independent uniform random number in the interval [0,1] for each particle. Based on random numbers and preset dwell probability Based on the magnitude relationship, two differentiated velocity updates are selected: if the random number is less than the dwell probability, a dwell behavior velocity update is performed; otherwise, a jump-type particle velocity update is performed. Then, the particle position is updated based on the updated particle velocity. The formula used is as follows:

[0103] ;

[0104] ;

[0105] In the formula, Indicates the current iteration number. Indicates the first Inertia weights in the next iteration It indicates that the i-th particle is in the... Speed ​​in the next iteration It indicates that the i-th particle is in the... Speed ​​in the next iteration This represents the local optimal position of an individual particle. , and Represents a random number in the range [0,1]. It indicates that the i-th particle is in the... Position in the next iteration It indicates that the i-th particle is in the... Position in the next iteration and These represent individual learning factors and group learning factors, respectively. Let d represent the Villevian flight step size vector, where d represents the dimension of the solution space of the optimization problem, corresponding to the number of variables to be optimized. This represents generating a row vector of 1 row and d columns with all elements being 1. This vector is used to match the dimensions of the solution space, so that random perturbations are applied evenly to each dimension of the optimization variable, ensuring that the magnitude of the search perturbations in each dimension is consistent and the search direction is balanced.

[0106] The particle search terminates by first updating the optimal particle position, and then terminating the search and obtaining the global optimal particle position when the particle fitness value is less than the fitness threshold or when the maximum number of iterations is reached; the global optimal particle position specifically refers to the optimal set of task node time parameters.

[0107] The optimal position update of the particle specifically involves re-evaluating the fitness value of all updated particles and comparing it with the current global optimal position of the particle based on the current particle's fitness value. If the current particle's fitness value is better, then the global optimal position of the particle is updated.

[0108] The task node dynamic time window reconstruction is used to redetermine the executable construction time range of each incomplete task node based on the effective optimal task node time parameter set, thereby improving the dynamic adaptability of the project schedule. Specifically, the start and end times of each task node in the optimal task node time parameter set are used as the recommended start and end times of each incomplete task node, respectively, to generate a dynamic time window for each incomplete task node. The dynamic time window includes the recommended start and end times.

[0109] The project schedule update is used to apply the optimized task node time parameters to the original project schedule, thereby enabling dynamic adjustment of the decoration design project schedule. Specifically, based on the dynamic time windows of each incomplete task node, the recommended start time and recommended end time of each incomplete task node are used to replace the planned start time and planned end time of the corresponding task node in the original project schedule, while keeping the actual completion time of completed task nodes and the actual start time of nodes under construction unchanged, resulting in the updated decoration design project schedule.

[0110] By performing the above operations, this invention addresses the technical problems of existing methods for optimizing project schedules, which often rely on fixed rules and manual experience to adjust schedule node times. These methods result in unreasonable schedule adjustments and fail to effectively reduce the overall project delay risk, leading to chaotic and inefficient project schedule management. The invention proposes an improved task node time parameter optimal search method based on a hybrid mutation of dual-mode velocity updates and particle global optimal positions. This method enhances the global optimization capability and local adjustment accuracy of task node time parameter searches, strengthens the adaptability to complex decoration construction constraints and dynamic schedule changes, and improves the rationality, stability, and executability of the updated project schedule. It achieves adaptive optimization of the decoration design project schedule with the goal of minimizing the overall delay days.

[0111] Example 5, see Figure 1 This embodiment is based on the above embodiment. Specifically, the intelligent project progress management module first calls the project progress risk prediction module to perform real-time prediction of the project progress risk of the current decoration design project, and obtains the overall real-time delay days of the current decoration design project. Based on the overall real-time delay days, it is determined whether the current decoration design project has a delay risk. When it is determined that the current decoration design project has a delay risk, the updated decoration design project progress plan is obtained through the project task node time optimization module. Construction progress management is carried out according to the project progress plan. Otherwise, the current decoration design project progress plan is maintained, thereby realizing intelligent management of the decoration design project progress.

[0112] The determination of whether the current decoration design project is at risk of delay is as follows: if the overall real-time delay days of the project are greater than 0, then the current decoration design project is at risk of delay, and the project task node time optimization module is triggered to dynamically optimize the start and end times of each unfinished task node to obtain an updated decoration design project schedule. If the overall real-time delay days of the project are less than or equal to 0, then the current decoration design project is expected to be completed on schedule or ahead of schedule, and the current decoration design project is at risk of delay. The project task node time optimization module is not triggered, and the current decoration design project schedule is maintained.

[0113] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0114] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

[0115] The present invention and its embodiments have been described above. This description is not restrictive, and the accompanying drawings are only one embodiment of the present invention; the actual structure is not limited thereto. In conclusion, if those skilled in the art are inspired by this description and design similar structures and embodiments without departing from the spirit of the invention, such designs should fall within the protection scope of the present invention.

Claims

1. A decoration design project progress management system based on big data, characterized by: It includes a standardized data acquisition module, a project progress risk prediction module, a project task node time optimization module, and a project progress intelligent management module; The standardized data acquisition module specifically obtains standardized data for decoration design project progress management through data collection and data preprocessing. The project progress risk prediction module is used to predict the overall delay risk of the entire decoration project and provide a basis for decision-making for task node time optimization. Specifically, it first constructs a project progress risk prediction model that can capture the high-order dependencies between task nodes and the dynamic changes of time series, trains the project progress risk prediction model, and finally performs real-time prediction of project progress risk to obtain the overall real-time delay days of the project. The project task node time optimization module is used to dynamically adjust the start and end times of each unfinished task node in the project, so as to realize adaptive optimization of the decoration design project schedule. Specifically, it includes filtering unfinished task nodes, obtaining a set of time parameters of nodes to be optimized, performing optimal search for task node time parameters based on an optimization algorithm improved by dual-mode velocity update and particle global optimal position hybrid mutation, reconstructing task node dynamic time windows and updating the project schedule, thereby obtaining the updated decoration design project schedule. The intelligent project progress management module first performs real-time prediction of project progress risks, obtains the total real-time delay days of the current decoration design project, and judges the delay risk based on the total real-time delay days. If there is a delay risk, the updated decoration design project progress plan is obtained through the project task node time optimization module; otherwise, the original progress plan is maintained, thereby realizing intelligent management of the decoration design project progress.

2. The decoration design project progress management system based on big data as described in claim 1, characterized in that: The project schedule risk prediction module specifically includes the following steps: Construct a project schedule risk prediction model; The training of the project schedule risk prediction model involves using standardized historical project schedule risk prediction data as model training data to iteratively train the prediction model and obtain the trained project schedule risk prediction model. Real-time prediction of project schedule risk involves inputting standardized real-time project schedule risk prediction data into a trained project schedule risk prediction model to obtain the overall real-time delay days for the project.

3. The decoration design project progress management system based on big data according to claim 2, characterized in that: The construction project schedule risk prediction model specifically includes the following steps: Constructing a task node graph involves treating each engineering task node as a node in the graph, building node features for each node, constructing an edge set and a node-edge adjacency matrix, and obtaining the task node graph through the construction of node features and adjacency matrix. The project task node refers to an independent construction task unit formed by decomposing the decoration design project according to the construction process, construction area, type of work and construction sequence. The edge connection rules in the edge set include the dependencies between task nodes, conflicts in construction areas, and the connection between work processes. The high-order interaction feature extraction specifically involves inputting the task node graph into a graph neural network and using graph convolution operations to aggregate the node features to obtain the high-order aggregated features of each node. Temporal augmentation feature extraction specifically involves arranging the high-order aggregated features of each node according to the time series to form an input sequence, which is then input into a bidirectional long short-term memory network for forward and reverse temporal modeling to obtain the temporal augmentation features of each node. The project schedule risk prediction output specifically involves average pooling the features of all nodes at the last time step to obtain the global feature vector of the overall project schedule, and then inputting it into a multilayer perceptron regression network to output the overall delay days of the project.

4. The decoration design project progress management system based on big data as described in claim 1, characterized in that: The project task node time optimization module specifically includes the following steps: The incomplete task node filtering is used to identify the construction task nodes that have not yet been completed in the decoration design project, and to provide adjustment objects for task node time optimization; specifically, based on the task node status identification results, the nodes under construction and the nodes not under construction are filtered out from all task nodes to form a set of incomplete task nodes. Obtain the set of time parameters for the task nodes to be optimized. Specifically, based on the set of incomplete task nodes, take the start time and end time of each incomplete task node as the parameters to be optimized to obtain the set of time parameters for the task nodes to be optimized. Optimal search for task node time parameters; The task node dynamic time window reconstruction specifically involves using the start and end times of each task node in the optimal task node time parameter set as the recommended start and end times for each incomplete task node, thereby generating a dynamic time window for each incomplete task node. The project schedule is updated by replacing the planned start and end times of the corresponding task nodes in the original project schedule with the recommended start and end times of each unfinished task node based on the dynamic time windows of each unfinished task node, while keeping the actual completion time of completed task nodes and the actual start time of nodes under construction unchanged, thus obtaining the updated decoration design project schedule.

5. The decoration design project progress management system based on big data according to claim 4, characterized in that: The optimal search for the task node time parameters specifically includes the following steps: Initializing the search population involves encoding the set of time parameters of the task nodes to be optimized into the position vectors of the search individuals in the optimization algorithm, thereby completing the initialization of the particle population. The particle fitness value is calculated by replacing the start and end times of the corresponding task nodes in the real-time engineering progress risk prediction standardized data with the set of task node time parameters corresponding to the particle's position vector, obtaining candidate engineering progress risk prediction standardized data, inputting the candidate engineering progress risk prediction standardized data into the trained engineering progress risk prediction model, obtaining the overall project delay days corresponding to the particle, and using the overall project delay days as the fitness evaluation value of the particle. The global optimal position of the particle undergoes a hybrid mutation process. First, based on the ratio of the current iteration count to the maximum iteration count, adaptive weighting coefficients for two mutation modes are calculated. Then, Gaussian random numbers following a standard normal distribution and Cauchy random numbers following a standard Cauchy distribution are generated. Finally, the global optimal position of the particle is vector-superimposed with the adaptively weighted mutation perturbation term to obtain the mutated global optimal position of the particle. ; Particle update; The particle search terminates by first updating the optimal particle position, and then terminating the search and obtaining the global optimal particle position when the particle fitness value is less than the fitness threshold or when the maximum number of iterations is reached; the global optimal particle position specifically refers to the optimal set of task node time parameters. The optimal particle position update specifically involves re-evaluating the fitness value of all updated particles and comparing it with the current global optimal position based on the current particle's fitness value. If the current particle's fitness value is better, then the global optimal position of the particle is updated.

6. The decoration design project progress management system based on big data according to claim 5, characterized in that: The particle update specifically involves generating an independent uniform random number in the interval [0,1] for each particle. Based on random numbers and preset dwell probability Based on the magnitude relationship, two differentiated velocity updates are selected: if the random number is less than the dwell probability, a dwell behavior velocity update is performed; otherwise, a jump-type particle velocity update is performed. Then, the particle position is updated based on the updated particle velocity. The formula used is as follows: ； In the formula, Indicates the current iteration number. Indicates the first Inertia weights in the next iteration It indicates that the i-th particle is in the... Speed ​​in the next iteration It indicates that the i-th particle is in the... Speed ​​in the next iteration This represents the local optimal position of an individual particle. , and Represents a random number in the range [0,1]. It indicates that the i-th particle is in the... Position in the next iteration and These represent individual learning factors and group learning factors, respectively. Let d represent the Villevian flight step size vector, where d represents the dimension of the solution space of the optimization problem, corresponding to the number of variables to be optimized. This represents generating a 1-row vector with d columns and all elements being 1.

7. The decoration design project progress management system based on big data according to claim 1, characterized in that: The intelligent project progress management module first calls the project progress risk prediction module to perform real-time prediction of the project progress risk of the current decoration design project, and obtains the overall real-time delay days of the current decoration design project. Based on the overall real-time delay days, it determines whether the current decoration design project has a delay risk. When it is determined that the current decoration design project has a delay risk, the project task node time optimization module obtains an updated decoration design project progress plan, and construction progress management is carried out according to the project progress plan. Otherwise, the current decoration design project progress plan is maintained, thereby realizing intelligent management of the decoration design project progress.

8. The decoration design project progress management system based on big data according to claim 1, characterized in that: The standardized data acquisition module specifically collects raw data for decoration design project progress management and performs data preprocessing to obtain standardized data for decoration design project progress management. The standardized data for decoration design project progress management includes standardized data for historical project progress risk prediction and standardized data for real-time project progress risk prediction; the data preprocessing specifically involves cleaning, normalizing, and selecting data features from the original data for decoration design project progress management to obtain standardized data for decoration design project progress management.

Images