Building decoration curtain wall construction progress intelligent monitoring method and system

Abstract

The invention discloses an intelligent monitoring method and system for the construction progress of a building decoration curtain wall. The method comprises the steps that multi-source construction data of a construction site are collected through Internet of Things equipment and a high-definition camera; constructing a multi-dimensional tensor model and extracting data association features by applying a local reversible mapping algorithm to generate a standardized feature data set; performing parallel computing through shared memory hierarchical process mapping by utilizing the standardized feature data set, and outputting a project progress state evaluation result; constructing a decision diagram based on the project progress state evaluation result, and generating a construction progress deviation report; predicting a construction progress trend by adopting a sample optimal private regression algorithm and generating an optimization scheme; and visually displaying the construction progress prediction report and carrying out early warning. According to the invention, the problems of low data integration efficiency, delayed decision response and weak multi-dimensional data association analysis capability of traditional curtain wall construction monitoring are solved, and intelligent and efficient monitoring of the construction progress of the building decoration curtain wall is realized.

Description

Technical Field

[0001] The present invention relates to the technical field of building engineering construction management, and in particular to a method and system for intelligently monitoring the construction progress of a building decorative curtain wall. Background Art

[0002] As a crucial component of modern architecture, the construction progress management of decorative curtain walls directly impacts overall project quality and duration. As construction projects increase in scale and complexity, efficient management of curtain wall construction progress becomes increasingly crucial. Traditional curtain wall construction management relies primarily on manual record-keeping and empirical judgment, resulting in low efficiency and prone to human error.

[0003] Currently, common construction progress monitoring technologies on the market include BIM-based construction progress simulation and camera-based on-site monitoring systems. The former uses BIM to build a 3D model, linking the construction plan with the model to visualize construction progress; the latter uses cameras to collect on-site data to simply identify and record construction status.

[0004] More advanced technologies combine the Internet of Things (IoT) and computer vision to collect and analyze construction site data. This technology deploys multiple sensor nodes to collect site images, temperature, humidity, noise, and other data. Using computer vision algorithms, it identifies construction status, compares it with pre-set construction plans, and generates progress reports.

[0005] However, when processing data from large-scale curtain wall construction projects, this technology faces challenges such as low data integration efficiency, delayed decision-making responses, and weak capabilities for analyzing multidimensional data correlations. In particular, existing technologies struggle to accurately identify construction anomalies and issue timely warnings when faced with complex construction environments and changing conditions. Furthermore, they lack the ability to make intelligent decisions to optimize the allocation of construction resources. Summary of the Invention

[0006] The purpose of the present invention is to provide a method and system for intelligent monitoring of the construction progress of building decorative curtain walls, aiming to solve the problems existing in the prior art such as low data integration efficiency, delayed decision response, and weak multi-dimensional data correlation analysis capabilities, and to achieve intelligent and efficient monitoring of the construction progress of building decorative curtain walls.

[0007] To solve the above technical problems, the present invention provides an intelligent monitoring method for the construction progress of building decorative curtain walls, comprising: collecting environmental parameters, material status, personnel location and construction images of the construction site through Internet of Things devices and high-definition cameras to obtain multi-source construction data; constructing a multidimensional tensor model, and applying a local reversible mapping algorithm in the multidimensional tensor model to extract data association features of the multi-source construction data to generate a standardized feature data set; utilizing the standardized feature data set, performing parallel calculations through shared memory hierarchical process mapping, evaluating the construction status, and outputting a project progress status evaluation result; constructing a decision diagram based on the project progress status evaluation result, comparing and analyzing the actual progress with the planned progress, identifying delay risk points, and generating a construction progress deviation report; based on the construction progress deviation report, using a sample optimal private regression algorithm to predict the construction progress trend and generate an optimization plan to form a construction progress prediction report; displaying the construction progress prediction report through a visual interface, and issuing an early warning based on a preset threshold.

[0008] Optionally, IoT devices and high-definition cameras can be used to collect environmental parameters, material status, personnel locations, and construction images of the construction site to obtain multi-source construction data, including:

[0009] Deploy environmental monitoring sensors, material tracking tags, personnel location devices, and equipment status monitors at construction sites;

[0010] The collected raw data is compressed, noise filtered, and outlier detected to obtain pre-processed multi-source data streams;

[0011] The pre-processed multi-source data stream is stored according to time, location and data type to form the multi-source construction data.

[0012] Optionally, applying a local reversible mapping algorithm in the multidimensional tensor model to extract data association features of the multi-source construction data to generate a standardized feature data set includes:

[0013] Map different types of data into a unified tensor space and construct an initial multi-dimensional tensor representation;

[0014] Applying a local reversible mapping algorithm to the initial multidimensional tensor representation to extract internal correlation features between the multi-source construction data and generate a feature mapping matrix;

[0015] The feature mapping matrix is ​​used to perform standardization processing on the data and correct outliers, and a standardized feature data set is output.

[0016] Optionally, applying a local reversible mapping algorithm to the initial multidimensional tensor representation to extract internal correlation features between the multi-source construction data and generate a feature mapping matrix includes:

[0017] Based on the initial multidimensional tensor representation, calculating the Euclidean distance between data points and constructing an initial distance matrix;

[0018] Using the initial distance matrix, a set of nearest neighbor points is selected to establish a local neighbor graph structure;

[0019] Based on the local neighbor graph structure, calculating the local reconstruction weight of each data point to form a weight matrix;

[0020] Using the weight matrix, a global covariance matrix is ​​constructed to solve the eigenvalue decomposition problem;

[0021] Based on the eigenvalue decomposition result, selecting main eigenvectors to obtain a low-dimensional embedding representation;

[0022] A nonlinear transformation is performed on the low-dimensional embedding representation to generate a feature mapping matrix.

[0023] Optionally, the standardized feature data set is used to perform parallel computing through shared memory hierarchical process mapping to evaluate the construction status and output a project progress status evaluation result, including:

[0024] Partition data and create indexes to form a data structure that supports parallel access;

[0025] Build a hierarchical process mapping framework to define the mapping relationship between data processing tasks and computing resources;

[0026] Use historical construction data to train construction period prediction models, resource consumption models, and quality assessment models, evaluate the current construction status, and output project progress status assessment results.

[0027] Optionally, a decision diagram is constructed based on the project progress status assessment results, the actual progress is compared and analyzed with the planned progress, delay risk points are identified, and a construction progress deviation report is generated, including:

[0028] Construct a decision diagram based on integrated tensor representation and establish a construction progress evaluation indicator system;

[0029] Matching the preset construction plan progress data with the decision diagram to establish a corresponding relationship matrix between the planned progress and the actual progress;

[0030] The progress difference is calculated through the corresponding relationship matrix, key delay points and influencing factors are identified, and a construction progress deviation report is generated.

[0031] Optionally, based on the construction progress deviation report, a sample optimal private regression algorithm is used to predict the construction progress trend and generate an optimization plan to form a construction progress prediction report, including:

[0032] Combine historical construction data to construct a sample data space and determine the prediction feature dimensions;

[0033] Applying a polynomial-time sample-optimal private regression algorithm in the sample data space to train a progress prediction model;

[0034] The progress prediction model is used to simulate progress changes under different adjustment strategies, select the optimal solution, and generate a construction progress prediction report.

[0035] Optionally, applying a polynomial time sample optimal private regression algorithm in the sample data space to train a progress prediction model includes:

[0036] Based on the sample data space, a time window matrix is ​​constructed, and the historical construction data is segmented to obtain a training sample set;

[0037] Using the training sample set, calculating the similarity weights between samples and constructing a weighted distance matrix;

[0038] Based on the weighted distance matrix, applying a kernel function to perform feature mapping to obtain a high-dimensional feature space;

[0039] In the high-dimensional feature space, iteratively optimizing regression parameters by gradient descent method to construct a private regression model;

[0040] Cross-validating the private regression model, selecting an optimal model parameter combination, and obtaining a validated regression model;

[0041] The verified regression model is applied to the current construction data to output the progress prediction result.

[0042] Optionally, after generating the standardized feature dataset, the following steps are also included:

[0043] Performing dimensionality reduction and information compression on the standardized feature dataset, including:

[0044] Based on the standardized feature dataset, applying a tensor decomposition algorithm to separate high-dimensional features;

[0045] Perform principal component analysis on the separated high-dimensional features and select the principal components whose feature contribution rate exceeds the preset threshold;

[0046] The feature data is reconstructed based on the selected principal components to obtain the key feature set after dimensionality reduction.

[0047] An intelligent monitoring system for the construction progress of a building decorative curtain wall comprises: a data acquisition module for collecting environmental parameters, material status, personnel location and construction images of a construction site through an Internet of Things device and a high-definition camera to obtain multi-source construction data; a data processing module for constructing a multidimensional tensor model for the multi-source construction data and applying a local reversible mapping algorithm to extract data association features to generate a standardized feature data set; a parallel computing module for utilizing the standardized feature data set to perform parallel computing through shared memory hierarchical process mapping, evaluate the construction status, and output a project progress status evaluation result; a progress analysis module for constructing a decision diagram based on the project progress status evaluation result, comparing and analyzing the actual progress with the planned progress, identifying delay risk points, and generating a construction progress deviation report; a prediction and optimization module for predicting the construction progress trend and generating an optimization plan based on the construction progress deviation report using a sample optimal private regression algorithm to form a construction progress prediction report; and a display and warning module for displaying the construction progress prediction report through a visual interface and issuing warnings based on preset thresholds.

[0048] The beneficial effects of the present invention are:

[0049] By constructing a multidimensional tensor model and applying a local reversible mapping algorithm, efficient processing and feature extraction of multi-source heterogeneous data are achieved, improving data integration efficiency.

[0050] The use of shared memory hierarchical process mapping technology for parallel computing has accelerated the processing of large-scale construction data and reduced decision response delays;

[0051] The construction progress analysis method based on compact decision diagram enhances the correlation analysis capability of multi-dimensional data and improves the accuracy of construction anomaly identification;

[0052] The construction progress is predicted by the sample-optimal private regression algorithm, providing a scientific optimization solution and strengthening the intelligent decision-making ability of construction resources;

[0053] A visual interface and early warning mechanism were designed to achieve intuitive display of construction progress and timely early warning of risks. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present disclosure, the following briefly introduces the drawings required for use in the embodiments. The drawings herein are incorporated into and constitute a part of the specification. These drawings illustrate embodiments consistent with the present disclosure and, together with the specification, are used to illustrate the technical solutions of the present disclosure. It should be understood that the following drawings only illustrate certain embodiments of the present disclosure and should not be regarded as limiting the scope. For those of ordinary skill in the art, other relevant drawings can be obtained based on these drawings without inventive effort.

[0055] Figure 1 This is a flow chart of the intelligent monitoring method for the construction progress of a building decoration curtain wall according to the present invention;

[0056] Figure 2 This is a flowchart of multi-source construction data collection of the present invention;

[0057] Figure 3 This is a flow chart of data preprocessing based on integrated tensor and local reversible mapping of the present invention;

[0058] Figure 4 This is a flow chart of the local reversible mapping algorithm of the present invention;

[0059] Figure 5 This is a structural diagram of the intelligent monitoring system for the construction progress of building decorative curtain walls of the present invention. DETAILED DESCRIPTION

[0060] In order to make the purpose, technical solutions and advantages of the embodiments of the present disclosure clearer, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only part of the embodiments of the present disclosure, not all of the embodiments. The components of the embodiments of the present disclosure generally described and shown in the drawings herein can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present disclosure provided in the drawings is not intended to limit the scope of the disclosure for which protection is sought, but merely represents selected embodiments of the present disclosure. Based on the embodiments of the present disclosure, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of the present disclosure.

[0061] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings.

[0062] The term "and / or" herein simply describes an association relationship, indicating that three relationships can exist. For example, A and / or B can represent the existence of A alone, the simultaneous existence of A and B, and the existence of B alone. In addition, the term "at least one" herein refers to any combination of at least two of any one or more of a plurality of items. For example, "at least one of A, B, and C" can represent any one or more elements selected from the set consisting of A, B, and C.

[0063] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of them. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0064] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0065] like Figure 1 As shown, the present invention provides a method for intelligently monitoring the construction progress of a building decorative curtain wall, comprising the following steps:

[0066] Step 101: Collect environmental parameters, material status, personnel location, and construction images of the construction site using IoT devices and high-definition cameras to obtain multi-source construction data;

[0067] First, an IoT perception layer, including environmental monitoring sensors, material tracking tags, personnel location devices, and equipment status monitors, is deployed at the construction site to form a multi-dimensional data collection network covering the entire construction process. High-definition camera arrays and multimodal sensor access systems are then configured, and unified data collection protocols and transmission standards are established to enable the simultaneous collection and transmission of multi-source data. Preliminary processing, including data compression, noise filtering, and outlier detection, is then performed at the edge computing node, outputting a preprocessed multi-source data stream. Finally, the preprocessed multi-source data stream is securely transmitted to a cloud data center via a 5G / WiFi network, where it is structured and stored according to time, location, and data type, forming the original dataset.

[0068] Step 102: constructing a multidimensional tensor model, and applying a local reversible mapping algorithm in the multidimensional tensor model to extract data association features of the multi-source construction data to generate a standardized feature data set;

[0069] In this step, a multidimensional tensor model is first constructed based on the original dataset, mapping different types of data (images, environmental parameters, location information, etc.) into a unified tensor space to form an initial multidimensional tensor representation. A local reversible mapping algorithm is then applied to the initial multidimensional tensor representation. The initial distance matrix is ​​constructed by calculating the Euclidean distance between data points. A set of nearest neighbors is selected to establish a local neighbor graph structure. Local reconstruction weights are calculated to form a weight matrix. A global covariance matrix is ​​constructed to solve the eigenvalue decomposition problem. Finally, the internal correlation features and spatiotemporal evolution patterns between the data are extracted to generate a feature mapping matrix. The feature mapping matrix is ​​then used to standardize the data and correct for outliers, ensuring comparability across data sources and scales. A standardized feature dataset is then output.

[0070] Step 103: Using the standardized feature data set, parallel computing is performed through shared memory hierarchical process mapping to evaluate the construction status and output a project progress status evaluation result;

[0071] First, based on a standardized feature dataset, a multi-level shared memory structure is designed to partition data for storage and establish efficient indexing, creating a data structure that supports parallel access. A hierarchical process mapping framework is then constructed to define the mapping relationship between data processing tasks and computing resources, establishing a parallel computing model. Historical construction data is then used to train a curtain wall construction progress analysis model, including a duration prediction model, a resource consumption model, and a quality assessment model, forming a model family. Finally, the trained model family is used to perform a multi-dimensional assessment of the current construction status, including construction progress, resource utilization efficiency, and construction quality, and output the project progress status assessment results.

[0072] Step 104: Construct a decision diagram based on the project progress status assessment results, compare and analyze the actual progress with the planned progress, identify delay risk points, and generate a construction progress deviation report;

[0073] In this step, a compact decision diagram using integrated tensor representation is first constructed based on the project progress status assessment results. This system establishes a construction progress assessment indicator system and a decision node relationship network. Pre-set construction plan progress data is then imported into the system and matched with the compact decision diagram to establish a correspondence matrix between the planned and actual progress. The correspondence matrix is ​​then used to calculate the difference between the actual and planned progress. Path analysis within the decision diagram is then used to identify key delay points and influencing factors, generating a progress variance analysis report. Finally, a risk assessment algorithm is applied to identify potential delay risk points and quantify their impact, ultimately producing a construction progress deviation report.

[0074] Step 105: Based on the construction progress deviation report, a sample optimal private regression algorithm is used to predict the construction progress trend and generate an optimization plan to form a construction progress prediction report;

[0075] Based on construction progress deviation reports and combined with historical construction data, a sample data space is constructed to determine the prediction feature dimensions and target variables. Then, within this sample data space, a polynomial time sample optimal private regression algorithm is applied. By constructing a time window matrix, calculating sample similarity weights, applying kernel function feature mapping, and optimizing regression parameters, a progress prediction model is trained to accurately predict the construction progress. Next, based on the trained private regression model, progress changes under different adjustment strategies are simulated, multiple possible construction plans are generated, and their effectiveness is compared. Finally, the optimal plan is selected, taking into account factors such as time, cost, and quality. Specific implementation steps are generated, and a construction progress forecast report and optimization plan are output.

[0076] Step 106: Display the construction progress forecast report through a visual interface and issue an early warning based on a preset threshold.

[0077] In this step, a multidimensional visualization interface, including modules such as timeline, progress comparison, resource allocation, and risk warning, was first designed to display construction progress forecast reports and optimization plans. A real-time linkage mechanism was then established between data processing results and the visualization interface to ensure the timeliness and accuracy of displayed data. Multi-level warning thresholds and trigger conditions were then set for factors such as schedule delays, resource shortages, and quality risks, establishing a hierarchical warning mechanism. Finally, monitoring data was analyzed in real time, automatically triggering alerts when warning conditions were met. Specific intervention recommendations were then generated based on the optimization plan, enabling intelligent monitoring of construction progress and risk warnings.

[0078] Furthermore, if Figure 2 As shown, in the above step 101, environmental parameters, material status, personnel location, and construction images of the construction site are collected through IoT devices and high-definition cameras to obtain multi-source construction data, including:

[0079] Step 201: Deploy environmental monitoring sensors, material tracking tags, personnel location equipment, and equipment status monitors at the construction site;

[0080] Step 202: compressing, noise filtering, and outlier detection are performed on the collected raw data to obtain a pre-processed multi-source data stream;

[0081] Step 203: The pre-processed multi-source data stream is stored according to time, location and data type to form the multi-source construction data.

[0082] Based on the specific conditions of the construction site, an IoT perception layer was designed and deployed, including environmental monitoring sensors, material tracking tags, personnel location devices, and equipment status monitors, forming a multi-dimensional data collection network covering the entire construction process. A high-definition camera array and multimodal sensor access system were configured, and unified data collection protocols and transmission standards were established to achieve the simultaneous collection and transmission of multi-source data. Preliminary processing, including data compression, noise filtering, and outlier detection, was performed at the edge computing node, outputting preprocessed multi-source data streams. These preprocessed multi-source data streams were securely transmitted to a cloud data center via a 5G / WiFi network, where they were structured and stored according to time, location, and data type to form the original data set.

[0083] Furthermore, if Figure 3 As shown, in the above step 102, the local reversible mapping algorithm is applied in the multi-dimensional tensor model to extract the data association features of the multi-source construction data to generate a standardized feature data set, including:

[0084] Step 301: Map different types of data into a unified tensor space to construct an initial multi-dimensional tensor representation.

[0085] The different types of data in the original dataset are preprocessed, including data cleaning, format unification, and missing value processing. Then, based on the time, space, and attribute dimensions of the data, a multidimensional coordinate system is constructed to determine the mapping rules for each type of data in the tensor space. Environmental parameters (such as temperature, humidity, wind speed, etc.) are then mapped to one slice of the multidimensional tensor, material status data (such as location, usage, etc.) is mapped to another slice, personnel location data is mapped to a third slice, and so on. Finally, these slices are integrated into a unified multidimensional tensor structure to form an initial multidimensional tensor representation that retains the multidimensional information and inherent relationships of the original data.

[0086] Step 302: Apply a local reversible mapping algorithm to the initial multi-dimensional tensor representation to extract internal correlation features between the multi-source construction data and generate a feature mapping matrix.

[0087] The detailed execution process of this step is as follows Figure 4 As shown in the figure, through a series of mathematical operations and transformations, the intrinsic correlation features between data are extracted from the initial multi-dimensional tensor representation, and a feature mapping matrix that can characterize these correlation relationships is generated.

[0088] Step 303: using the feature mapping matrix to perform data standardization and outlier correction, and output a standardized feature data set.

[0089] Based on the feature mapping matrix, data standardization parameters, including the mean vector and standard deviation vector, are calculated. The Z-score standardization method is then applied to the raw data, converting data of different scales and units to the same standard for easy comparison. The data relationships implicit in the feature mapping matrix are then used to identify potential outliers, which are then corrected using median substitution or local interpolation. Finally, the standardized data is integrated to form a standardized feature dataset with a uniform scale, less noise, and a more defined feature structure, laying the foundation for subsequent parallel computing and construction status assessment.

[0090] Furthermore, if Figure 4 As shown, in the above step 302, a local reversible mapping algorithm is applied to the initial multi-dimensional tensor representation to extract the internal correlation features between the multi-source construction data and generate a feature mapping matrix, including:

[0091] like Figure 4 As shown, in step 302, a local reversible mapping algorithm is applied to the initial multidimensional tensor representation to extract internal correlation features between the multi-source construction data and generate a feature mapping matrix, which includes the following detailed steps:

[0092] Step 401: Based on the initial multidimensional tensor representation, calculate the Euclidean distance between data points and construct an initial distance matrix. The initial multidimensional tensor representation is converted into a vector set, each vector representing the position of a data point in the multidimensional feature space. Then, for each pair of data points (x i ,x j ), calculate the Euclidean distance d(x i ,x j )=sqrt(∑(x i k -x j k ) 2 ), where k represents the feature dimension. The distance calculation results between all pairs of data points are then organized into an N×N matrix (N is the number of data points), forming the initial distance matrix. This matrix reflects the relative positional relationships of the data points in the original feature space and provides a basis for subsequent local structure analysis.

[0093] Step 402: Using the initial distance matrix, select the nearest neighbor point set and establish a local neighbor graph structure. Based on the initial distance matrix, for each data point x i Determine the K nearest neighbor points (K is a preset parameter, usually 5 to 20) between these points and the center point x i The distance is minimized. Then, using data points as nodes and nearest neighbor relationships as edges, a local neighbor graph G = (V, E) is constructed, where V is the node set and E is the edge set. If point j is one of point i's K nearest neighbors, an edge is established between i and j. Finally, each edge in the neighbor graph is assigned a weight, typically based on the inverse of the distance or a Gaussian kernel function, to form a complete weighted local neighbor graph structure. This structure preserves the topological relationships of the data within the local area, providing a basis for subsequent local reconstruction.

[0094] Step 403: Based on the local neighbor graph structure, calculate the local reconstruction weight of each data point to form a weight matrix. Assume that each data point x i It can be reconstructed by the linear combination of its K nearest neighbors, that is, x i ≈∑w ij ·x j , where j traverses x i All neighboring points of w ij is the reconstruction weight. Then, by minimizing the reconstruction error ∑|x i -∑w ij ·x j | 2 To solve the optimal reconstruction weight, while satisfying ∑w ij= 1. The above process is then repeated for each data point to calculate the reconstruction weights for all data points. Finally, these weights are organized into an N×N sparse matrix (most elements are 0, and only the positions corresponding to the nearest neighbors have non-zero values), forming the weight matrix W. This matrix captures the linear relationships between the data within the local neighborhood and reflects the degree of interdependence between data points.

[0095] Step 404: Using the weight matrix, construct a global covariance matrix and solve the eigenvalue decomposition problem.

[0096] Based on the weight matrix W, the matrix M is calculated as (IW) T (IW), where I is the identity matrix. Then calculate the eigenvalues ​​and eigenvectors of the matrix M, that is, solve the equation Mv=λv, where λ is the eigenvalue and v is the corresponding eigenvector. Then sort the eigenvalues ​​to determine the intrinsic dimension d of the data (usually determined by analyzing the distribution of eigenvalues ​​or setting a cumulative contribution rate threshold). Finally, retain the eigenvectors corresponding to the smallest d+1 non-zero eigenvalues ​​(ignoring the eigenvectors corresponding to eigenvalues ​​of 0) to form an eigenvector matrix. This step reveals the global geometric characteristics and intrinsic dimensions of the data by analyzing the structure of the weight matrix.

[0097] Step 405: Based on the eigenvalue decomposition result, select the main eigenvectors to obtain a low-dimensional embedding representation.

[0098] Select d eigenvectors corresponding to the 2nd to d+1th smallest eigenvalues ​​from the eigenvector matrix (ignoring the eigenvector corresponding to the smallest eigenvalue). These d eigenvectors are then used as column vectors to form an N×d matrix Y. Each row of the matrix Y is then normalized to ensure that the representation of each data point in the low-dimensional space has a unit norm. Finally, the processed matrix is ​​used as the embedded representation of the original data in the low-dimensional space, with each row corresponding to the coordinates of a data point in the d-dimensional space. This low-dimensional embedded representation retains the local neighborhood structure and global geometric characteristics of the original data while reducing the complexity of the data.

[0099] Step 406: Perform a nonlinear transformation on the low-dimensional embedding representation to generate a feature mapping matrix. Select an appropriate nonlinear transformation function f(·), such as a sigmoid function, a tanh function, or a ReLU function, to enhance the expressive power of the low-dimensional embedding representation. Then apply a nonlinear transformation to each element of the low-dimensional embedding representation to obtain a transformed representation Z=f(Y). The transformed representation is then scaled and translated so that its elements are distributed within a suitable range. Finally, the transformed matrix is ​​combined with the original features to form a feature mapping matrix Φ. This feature mapping matrix not only contains the geometric information after dimensionality reduction, but also enhances the expressive power of complex patterns through nonlinear transformations, providing a powerful tool for subsequent data standardization and outlier correction.

[0100] Furthermore, in the above step 103, the standardized feature data set is used to perform parallel computing through shared memory hierarchical process mapping to evaluate the construction status and output the project progress status evaluation result, including:

[0101] Step 501: Partition the data and store it in indexes to form a data structure that supports parallel access. In this step, the standardized feature data set is first partitioned according to the time dimension, space dimension, and feature dimension to form multiple data blocks. A dedicated memory area is then designed for each data block, and corresponding resources are allocated in the shared memory space. A multi-level index structure is then established, including block-level indexes (to identify different data blocks), record-level indexes (to locate specific records within a block), and feature-level indexes (to quickly access specific features), to achieve efficient positioning and retrieval of data. Finally, a parallel access control mechanism is created, including read-write locks and atomic operation support, to ensure data consistency and integrity during concurrent access by multiple processes. This data structure design significantly improves data access speed, reduces memory contention and data transmission overhead during parallel computing, and lays the foundation for subsequent high-performance computing.

[0102] Step 502: Construct a hierarchical process mapping framework to define the mapping relationship between data processing tasks and computing resources. In this step, a three-layer process architecture is first designed, including a master control layer (responsible for task scheduling and resource allocation), a computing layer (performing specific data processing tasks), and a communication layer (coordinating data exchange between processes). Then, based on the data partition structure, the construction progress analysis task is decomposed into multiple subtasks, each of which is responsible for processing a specific data block or performing a specific computing step. Next, a task-resource mapping table is established to reasonably allocate computing resources such as CPU and memory based on the computing intensity and data dependencies of the task, taking load balancing factors into consideration. Finally, a dynamic task scheduling mechanism is implemented, which can adjust the task execution order and resource allocation strategy in real time according to the system load status and task priority. This hierarchical process mapping framework effectively solves the parallel efficiency problem in large-scale data processing and improves the system's throughput and responsiveness.

[0103] Step 503: Use historical construction data to train a construction period prediction model, a resource consumption model, and a quality assessment model to evaluate the current construction status and output a project progress status assessment result. In this step, the construction period prediction model (for predicting the completion time of each construction task), the resource consumption model (for estimating the use of manpower, materials, and equipment), and the quality assessment model (for assessing construction quality and potential risks) are first trained based on the historical construction data stored in shared memory. The currently collected real-time data is then input into these trained models through a parallel computing framework to obtain construction period prediction results, resource consumption results, and quality assessment scores, respectively. These three assessment results are then integrated to generate multi-dimensional project progress status assessment indicators, including time progress indicators, resource utilization indicators, and quality compliance indicators. Finally, these assessment indicators are compared with the preset schedule plan and standards to generate a project progress status assessment result that includes schedule deviations, resource anomalies, and quality issues. This assessment result comprehensively reflects the current construction status and provides a data foundation for subsequent decision-making and analysis.

[0104] For example, during the training of the construction period prediction model, we first collected historical construction data from 20 curtain wall projects of similar scale, totaling approximately 150,000 records, including core information such as the planned start time, actual start time, planned completion time, and actual completion time of each construction task. The training used the random forest regression algorithm, and the key parameter settings included: the number of decision trees ntrees = 100, the maximum tree depth max depth =10, feature sampling ratio feature fraction =0.8, sample sampling ratio bagging fraction= 0.7. Input features include 15 key features, including construction task type (categorized into eight categories, such as frame installation, glass installation, and sealing), task duration estimate, completion status of prerequisite tasks, resource input, and weather conditions (classified on a scale of 0-5). The final model achieved a prediction accuracy of 1.2 days (mean absolute error) and 1.8 days (root mean square error) on the validation set.

[0105] The resource consumption model uses the gradient boosting decision tree (GBDT) algorithm. The training parameters include: learning rate rate =0.05, number of iterations n estimators =500, maximum tree depth max depth =6, minimum number of split samples min samplessplit = 20. The input features include 12 features such as construction task type, area size, curtain wall type (divided into 6 categories), construction worker experience level (level 1-5), and equipment status index. The target variable is the daily consumption of various resources, including labor hours, material usage, and equipment usage time. The performance indicators of the model on the validation set are: labor hours prediction R 2 =0.86, material consumption prediction R 2 =0.83, equipment usage time prediction R 2 =0.79.

[0106] The quality assessment model uses an integrated method of support vector machine (SVM) and multi-layer perceptron (MLP). The SVM parameters include: kernel function kernel = 'rbf', penalty coefficient C = 10.0, gamma = 0.01; the MLP parameters include: hidden layer structure hidden layersizes =(64,32,16), activation function activation = 'relu', initial learning rate learning rate = 0.001, and the L2 regularization coefficient alpha = 0.0001. The input features include 18 features, including construction process compliance scores, material inspection indicators, installation accuracy measurement data, and environmental condition impact indexes. The target variables are quality grade (classified as A, B, C, and D) and defect prediction probability. The ensemble model achieved a classification accuracy of 89.4% and an F1 score of 0.87 on the validation set.

[0107] A practical application example: When assessing the construction status of the east facade curtain wall of a large commercial building, the system first collected that day's construction data, including the area of ​​aluminum alloy frame installation completed, the number of glass panels installed, and the length of sealant application. The output of the construction period prediction model indicated that, based on the current progress, curtain wall installation in this area would be completed 4.2 days later than planned. The resource consumption model analysis revealed that the glass panel utilization rate was 7.8% higher than expected, potentially leading to material shortages later. The quality assessment model detected that the installation accuracy deviation in the northwest corner exceeded the standard by 1.2mm, resulting in a quality rating of B. The system synthesized these three results to generate a project progress status assessment report, identifying schedule delays, excessive material consumption, and quality risks, providing data support for subsequent decision-making.

[0108] Furthermore, in step 104, a decision diagram is constructed based on the project progress status assessment results, the actual progress is compared and analyzed with the planned progress, delay risk points are identified, and a construction progress deviation report is generated, including:

[0109] Step 601: Construct a decision graph using an integrated tensor representation and establish a construction progress evaluation indicator system. In this step, a multidimensional decision space is first designed based on the project progress status assessment results. This space dimension includes a time dimension (reflecting the progress plan), a spatial dimension (representing the construction area), and a feature dimension (containing various evaluation indicators). The evaluation indicators are then organized into a tensor structure, where each element of the tensor represents a specific indicator value at a specific time and in a specific area, forming an integrated tensor representation. A decision graph is then constructed based on this tensor structure, where nodes represent construction tasks or states, edges represent dependencies between tasks or state transition conditions, and edge weights reflect the importance or probability of the corresponding relationships. Finally, based on this decision graph, a comprehensive evaluation indicator system is established, including progress achievement rate, resource utilization efficiency, and quality qualification rate, with thresholds and weights set for each indicator. This decision graph structure using an integrated tensor representation effectively captures the complex relationships and multidimensional features in construction progress management, providing a structured analysis framework for accurately identifying problems.

[0110] Step 602: Match the preset construction plan progress data with the decision diagram to establish a correspondence matrix between the planned progress and the actual progress. In this step, the preset construction plan progress data is first imported, including the planned start time, planned completion time, planned resource input, and expected quality requirements for each construction task. This planned data is then converted into a plan tensor representation using the same organizational structure as the decision diagram. Then, through operations such as time alignment, spatial mapping, and feature matching, a correspondence is established between the plan tensor and the actual tensor, identifying the planned data point corresponding to each actual data point. Finally, based on this correspondence, a plan-actual correspondence matrix is ​​constructed, with each element of the matrix representing the mapping relationship between the planned value and the actual value along a specific dimension. This correspondence matrix places the planned progress and actual progress within the same analytical framework, allowing for direct comparison between the two and providing a foundation for the next step of variance analysis.

[0111] Step 603: Schedule variances are calculated using the correspondence matrix to identify key delay points and influencing factors, and a construction progress deviation report is generated. In this step, schedule variance values ​​are first calculated based on the correspondence matrix for various dimensions, including time variance (the difference between actual completion time and planned completion time), resource variance (the difference between actual resource consumption and planned resource input), and quality variance (the difference between actual quality level and expected quality requirements). Key delay points—construction tasks or areas with significant variances and wide-ranging impacts—are then identified based on features such as variance size, trend, and fluctuation. Path analysis and causal inference algorithms are then applied to extract the main factors contributing to delays from the decision graph, such as resource shortages, technical difficulties, weather conditions, or management issues. Finally, these analysis results are integrated into a structured construction progress deviation report, which includes an overall progress status, a list of key delayed tasks, an analysis of key influencing factors, and potential risk warnings. This deviation report not only identifies current progress issues but also reveals their causes and impact, providing a basis for targeted decision-making for subsequent progress optimization.

[0112] Furthermore, in step 105, based on the construction progress deviation report, a sample optimal private regression algorithm is used to predict the construction progress trend and generate an optimization plan to form a construction progress prediction report, including:

[0113] Step 701: Construct a sample data space based on historical construction data and determine the predictive feature dimensions. In this step, data from historical curtain wall construction projects similar to the current project is first collected, including information such as progress records, resource allocation, environmental conditions, and technical parameters. This historical data is then integrated with the analysis results from the construction progress deviation report to construct a comprehensive dataset encompassing multiple influencing factors. Next, through correlation analysis and feature importance assessment, key features that significantly impact the construction progress are selected from the numerous influencing factors, such as staffing ratios, material supply timeliness, equipment utilization efficiency, and weather condition impact index. Finally, the feature dimensions and target variables of the predictive model are determined to form a structured sample data space. This step not only integrates historical experience and current conditions but also reduces interference from irrelevant factors through scientific feature selection, laying the data foundation for building a highly accurate predictive model.

[0114] Step 702: Apply a polynomial-time sample-optimal private regression algorithm to the sample data space to train a progress prediction model. This step involves constructing a regression model capable of accurately predicting construction progress within the sample data space through a series of mathematical processing and machine learning techniques. This algorithm optimizes computational complexity to ensure model training is completed within polynomial time. A private regression mechanism protects sensitive data from direct exposure while maintaining prediction accuracy.

[0115] Step 703: Utilizing the progress prediction model, the project simulates progress changes under different adjustment strategies, selects the optimal solution, and generates a construction progress forecast report. In this step, various possible adjustment strategies are first designed based on the current construction status and identified delay risk points, such as increasing human resources, improving construction processes, adjusting the material supply chain, or rescheduling the construction sequence. The parameters of these adjustment strategies are then input into the trained progress prediction model to simulate construction progress changes over a future period (e.g., two weeks, one month, or the entire remaining construction period). The simulation results are then evaluated across multiple dimensions, including progress recovery, resource input efficiency, implementation difficulty, and potential risks. Finally, based on the comprehensive evaluation, the optimal adjustment solution is selected and a detailed construction progress forecast report is generated, including progress trend forecasts, estimated completion times for key nodes, resource demand forecasts, and specific implementation recommendations. This model-based solution evaluation and selection process significantly improves the scientific nature and reliability of decision-making and effectively reduces the risks associated with subjective judgment.

[0116] Furthermore, in step 702, applying a polynomial time sample optimal private regression algorithm in the sample data space to train a progress prediction model includes:

[0117] Step 801: Based on the sample data space, a time window matrix is ​​constructed and the historical construction data is segmented to obtain a training sample set. In this step, an appropriate time window size w (e.g., 3, 7, or 14 days) is first determined. This window is used to capture short-term variations in construction progress. The historical construction data is then segmented using a sliding window approach, with the data within each window forming a sample. The window sliding step can be set to 1 day or longer. For each sample, a feature vector x (containing the eigenvalues ​​at the start of the window) and a target vector y (representing the progress status at the end of the window) are constructed to form a "feature-target" pair. Finally, these sample pairs are organized into a structured training sample set D = {(x1, y1), (x2, y2), ..., (x, y)}, where n is the number of samples. This time window-based segmentation method effectively captures the temporal characteristics and variation patterns of construction progress, providing high-quality sample data for subsequent model training.

[0118] Step 802: Using the training sample set, calculate the similarity weights between samples and construct a weighted distance matrix. In this step, first design a similarity measurement function that is suitable for the characteristics of construction data. This function comprehensively considers the Euclidean distance of numerical features, the matching degree of category features, and the decay effect of time factors. Then, for each pair of samples (x i ,x j ) Calculate the similarity score s(x i ,x j ), which reflects the similarity between the two construction scenes. Then, the weight coefficient w between the samples is calculated based on the similarity score. ij , usually using the Gaussian kernel function w ij =exp(-d 2 (x i ,x j ) / σ 2 ), where d 2 Represents the square of the distance, σ is the kernel width parameter. Finally, the weight coefficients of all sample pairs are organized into an n×n weighted distance matrix W, and the elements w in this matrix are ij represents the influence weight of sample i on sample j. This weight calculation method, which takes sample similarity into account, can enhance the influence of similar scenarios in regression analysis and improve the model's adaptability to specific construction conditions.

[0119] Step 803: Based on the weighted distance matrix, apply the kernel function to perform feature mapping to obtain a high-dimensional feature space. In this step, first select the kernel function K(x i ,x j), such as radial basis kernel function (RBF), polynomial kernel function or custom composite kernel function. Then the kernel function is used to implicitly map the samples in the original feature space to the high-dimensional feature space, that is, This mapping enhances the model's ability to express complex nonlinear relationships. Then construct the kernel matrix K, where This matrix encodes the inner product relationship of samples in high-dimensional space. Finally, combined with the weighted distance matrix W, a weighted kernel matrix K is constructed that considers the similarity of samples. W , whose elements are K Wij =w ij ·K ij The application of this kernel method avoids the complexity of direct calculation in high-dimensional space (kernel trick) while retaining the powerful expression ability of high-dimensional space, which is suitable for processing complex nonlinear relationships in construction progress prediction.

[0120] Step 804: In the high-dimensional feature space, the regression parameters are iteratively optimized by the gradient descent method to construct a private regression model. In this step, first, based on the weighted kernel matrix K W , construct the objective function J(θ), which measures the weighted squared error between the predicted value and the actual value, and adds a regularization term to prevent overfitting. Then initialize the model parameters θ, including the coefficients of the kernel function and the bias term. Then apply stochastic gradient descent (SGD) or its variants (such as Adam, RMSprop, etc.) to iteratively update the parameters, that is, Where α is the learning rate, is the gradient of the objective function. At each iteration, the prediction error for the current parameters is calculated, and the parameter direction and step size are adjusted accordingly. Finally, the iteration stops when the preset number of iterations is reached or the objective function change is less than a threshold, and the final optimized model parameters θ* are output. This gradient descent-based optimization method efficiently finds the local optimal solution to the objective function while protecting the privacy of the original data through a private regression mechanism (using only the weighted kernel matrix instead of the original data).

[0121] Step 805: Cross-validate the private regression model, select the optimal model parameter combination, and obtain the verified regression model. In this step, first determine the set of hyperparameters that need to be optimized, such as the parameters of the kernel function, the regularization coefficient, the learning rate, etc. Then design the parameter search space, including the candidate value ranges of multiple parameters. Then, use the k-fold cross-validation method (usually k = 5 or 10) to randomly divide the training sample set into k subsets, use k-1 subsets to train the model in turn, and verify the performance with the remaining 1 subset. Repeat k times and take the average as the performance indicator of the parameter combination. Use algorithms such as grid search or Bayesian optimization to find the parameter combination that minimizes the verification error in the parameter search space. Finally, use the optimal parameter combination found to retrain the model on all training sample sets to obtain the final regression model after verification. This parameter optimization process can effectively avoid overfitting or underfitting of the model and improve the generalization performance of the model on unseen data.

[0122] Step 806: Apply the verified regression model to the current construction data and output the progress prediction result. In this step, the latest status data of the current construction project is first preprocessed in the same way as the training sample, and the corresponding feature vector x is extracted. current Then, x current Input the verified regression model and calculate the predicted value y pred =f(x current ; θ*), where f represents the regression function and θ* represents the optimized model parameters. The prediction confidence interval is then calculated, reflecting the reliability of the prediction results. Finally, the prediction results and confidence intervals are organized into a structured schedule forecast, including information such as expected completion time, schedule trends, and predictions for key milestones. This model prediction method, based on rigorous training and validation, provides reliable schedule forecasts and provides data support for construction management decisions.

[0123] Furthermore, after generating the standardized feature dataset in step 102, the following steps are further included:

[0124] Performing dimensionality reduction and information compression on the standardized feature dataset, including:

[0125] Step 901: Based on the standardized feature dataset, apply the tensor decomposition algorithm to separate high-dimensional features. In this step, the standardized feature dataset is first reorganized into a multi-dimensional tensor structure. Where n represents the order of the tensor, corresponding to the number of dimensions of the data (such as time, space, feature type, etc.). Then select a tensor decomposition algorithm that is suitable for the characteristics of curtain wall construction data. Commonly used ones include CP decomposition (CANDECOMP / PARAFAC decomposition) and Tucker decomposition. CP decomposition decomposes the original tensor X into the sum of multiple tensors of rank 1, that is, where λ r is the weight coefficient, a j r is the factor vector on the j-th dimension, represents a vector outer product operation; Tucker decomposition decomposes the original tensor into the product of a core tensor and a factor matrix, i.e., X≈G×1A1×2A2×...×A, where G is the core tensor, A is the factor matrix of the jth dimension, and × represents the tensor-matrix product along the jth dimension. The decomposition parameters are then solved using optimization algorithms such as alternating least squares (ALS) or gradient descent to minimize the reconstruction error. Finally, the factor vectors or factor matrices obtained from the decomposition are used to extract the underlying patterns and structural information from the original high-dimensional features. This tensor decomposition method can effectively handle the complex correlation structures of multidimensional data and provide a clearer feature representation for subsequent principal component analysis.

[0126] Step 902: Perform principal component analysis on the separated high-dimensional features and select the principal components whose feature contribution rate exceeds the preset threshold. In this step, the factor matrix or reconstructed features obtained by tensor decomposition are first organized into a two-dimensional matrix form X mat , rows represent samples, columns represent features. Then calculate the feature covariance matrix C = X mat T ·X mat / (n-1), where n is the number of samples. Then perform eigenvalue decomposition on the covariance matrix C and solve the characteristic equation Cv=λv to obtain the eigenvalues ​​{λ1,λ2,...,λ} (arranged in descending order) and the corresponding eigenvectors {v1,v2,...,v}, where p is the characteristic dimension. Calculate the contribution rate r of each eigenvalue i =λ i / ∑ j λ j and cumulative contribution rate R k =∑ {i＝1} k r i , the contribution rate reflects the ability of the corresponding principal component to explain the variability of the original data. Finally, the number of principal components k is determined so that the cumulative contribution rate R k The first time a pre-set threshold T (usually set at 85%, 90%, or 95%) is exceeded, the first k eigenvectors {v1, v2, ..., v} are selected as principal component basis vectors. This contribution-based principal component selection method significantly reduces data dimensionality while preserving the data's key information, thus reducing the computational burden for subsequent analysis.

[0127] Step 903: Reconstruct the feature data based on the selected principal components to obtain the key feature set after dimensionality reduction. In this step, the selected k principal component feature vectors are first combined into a projection matrix Then the original feature data Xmat Projecting to the principal component space to obtain the feature representation after dimensionality reduction Each row of Y represents the coordinates of a sample in the k-dimensional principal component space. The reduced features are then interpreted and named. By analyzing the relationship between the principal components and the original features, each principal component is given a physical or business meaning, such as "construction progress factor," "resource utilization factor," or "environmental impact factor." Finally, the reduced feature set is integrated with the necessary original features (such as timestamps, location identifiers, and other key index features) to form the final key feature set. This reduced feature set retains the main information of the original data while significantly reducing the data's dimensionality and redundancy, improving the efficiency of subsequent calculations and the stability of the model. Through the combined application of tensor decomposition and principal component analysis, effective compression of high-dimensional construction data and extraction of key information are achieved, providing a concise and information-rich feature representation for efficient analysis and prediction of construction progress.

[0128] In step 106, the construction progress forecast report is displayed through a visual interface, and an early warning is issued according to a preset threshold value, specifically including: designing a multi-dimensional visual interface including modules such as timeline, progress comparison, resource allocation and risk early warning; establishing a real-time linkage mechanism between data processing results and the visual interface to ensure the timeliness and accuracy of displayed data; setting multi-level early warning thresholds and trigger conditions such as progress delays, resource shortages, and quality risks, and establishing a hierarchical early warning mechanism; performing real-time analysis on monitoring data, automatically triggering an alarm when the early warning conditions are met, and generating specific intervention suggestions based on the optimization plan to achieve intelligent monitoring of construction progress and risk early warning.

[0129] In the curtain wall construction project of a high-rise office building, the system designed a multi-dimensional visual interface for progress monitoring and early warning. The timeline module uses a Gantt chart format, with the horizontal axis representing the date (from the start of the project to the expected end) and the vertical axis representing different construction areas (such as the 1st to 5th floors of the south facade, the 6th to 10th floors of the south facade, the 1st to 5th floors of the east facade, etc.). Each construction task is represented by a colored bar. The system displays the planned progress with a gray background and the actual progress with a color overlay to intuitively display the progress compliance. For example, the skeleton installation task of the 6th to 10th floors of the south facade was originally planned to be completed from August 10th to August 20th, but the actual progress bar showed that it was only 75% completed as of August 18th. The system automatically marked the task in yellow to indicate the potential risk of delay.

[0130] The progress comparison module features a "Planned vs. Actual" line chart that displays the planned completion percentage and actual completion percentage for each area. For example, the planned completion percentage for the north facade curtain wall installation was 65%, while the actual completion percentage was 58%. The system automatically calculated a deviation rate of -10.8% and indicated the degree of deviation using a gradient color: green indicates ahead of schedule or on schedule (deviation ≥ 0%), yellow indicates minor delays (deviation between -10% and 0%), orange indicates moderate delays (deviation between -20% and -10%), and red indicates severe delays (deviation < -20%).

[0131] The resource allocation module uses a combination of dashboards and heat maps to display the planned allocation, actual usage, and forecasted demand for various resources. For example, the human resources dashboard shows that the current demand for curtain wall installers is 42, with an actual allocation of 38, leaving a shortfall of 4. In the material heat map, aluminum extrusions appear green (sufficient inventory), while specialty sealants appear red (shortage, expected to run out in three days). Based on resource model predictions, the system recommends deploying six additional installers to the south facade construction area and ordering the next batch of sealant five days in advance to avoid delays caused by material shortages.

[0132] The risk warning module has a three-level alert mechanism: Level 1 (yellow, low risk) for tasks with delays of less than three days or resource shortages that haven't impacted the critical path; Level 2 (orange, medium risk) for tasks with delays of three to seven days or resource shortages that have impacted non-critical paths; and Level 3 (red, high risk) for tasks with delays exceeding seven days or resource shortages that have impacted the critical path. For example, during real-time monitoring on one particular day, the system detected that the west facade curtain wall installation was five days behind schedule. At the same time, the quality of the skewer installation in this area received a C rating (the standard is B or higher). This immediately triggered a Level 2 alert, automatically generating an alert and sending it to the project manager and quality supervisor. The alert included: a description of the problem (the west facade curtain wall installation was five days late and the quality was substandard), an impact assessment (this would delay subsequent waterproofing work and impact the overall project schedule by seven days), a cause analysis (deviations in skewer material specifications and insufficient installation worker skills), and recommended interventions (temporarily dispatching two senior technicians to make minor adjustments to the installed skewer and requesting expedited supply of a new batch of standard skewer materials).

[0133] Through this multi-dimensional visual interface and multi-level early warning mechanism, the system achieves an intuitive display of construction progress and timely early warning of risks. Project managers can quickly grasp the construction status, identify problems in a timely manner and take targeted measures, greatly improving the efficiency and accuracy of curtain wall construction progress management.

[0134] like Figure 5 As shown, the present invention also provides an intelligent monitoring system for the construction progress of a building decorative curtain wall, comprising:

[0135] Data acquisition module 1 is used to collect environmental parameters, material status, personnel location and construction images of the construction site through IoT devices and high-definition cameras to obtain multi-source construction data;

[0136] Data processing module 2, used to construct a multidimensional tensor model for the multi-source construction data and apply a local reversible mapping algorithm to extract data association features to generate a standardized feature data set;

[0137] A parallel computing module 3 is configured to utilize the standardized feature data set to perform parallel computing through shared memory hierarchical process mapping, evaluate the construction status, and output a project progress status evaluation result;

[0138] The progress analysis module 4 is used to construct a decision diagram based on the project progress status assessment results, compare and analyze the actual progress with the planned progress, identify delay risk points, and generate a construction progress deviation report;

[0139] Prediction and optimization module 5, for predicting the construction progress trend and generating an optimization plan based on the construction progress deviation report using a sample optimal private regression algorithm to form a construction progress prediction report;

[0140] The display and warning module 6 is used to display the construction progress forecast report through a visual interface and issue an early warning based on a preset threshold.

[0141] During the specific implementation process, the data acquisition module 1 is responsible for deploying Internet of Things devices and high-definition cameras at the construction site to collect multi-source data such as environmental parameters, material status, personnel location and construction images; the data processing module 2 is responsible for constructing a multidimensional tensor model, extracting data association features through a local reversible mapping algorithm, and generating a standardized feature data set; the parallel computing module 3 uses shared memory hierarchical process mapping technology to efficiently process the standardized feature data set in parallel to evaluate the current construction status; the progress analysis module 4 constructs a decision diagram based on the evaluation results, compares the actual progress with the planned progress, and identifies delay risk points; the prediction and optimization module 5 uses a sample optimal private regression algorithm to predict the construction progress trend and generate an optimization plan; the display and warning module 6 visualizes the prediction report and optimization plan, and issues a warning when the risk exceeds the threshold.

[0142] The present invention realizes intelligent and efficient monitoring of the construction progress of building decorative curtain walls through technical means such as multi-source data acquisition, data preprocessing based on integrated tensors and locally reversible mapping, parallel computing of shared memory hierarchical process mapping, construction progress analysis based on compact decision graphs, progress prediction of sample optimal private regression, and visual display and early warning. It effectively solves the problems of low data integration efficiency, decision response delay, and weak multi-dimensional data correlation analysis capabilities existing in traditional technologies.

[0143] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. All equivalent structural transformations made by using the contents of the present invention description and drawings under the concept of the present invention, or direct / indirect application in other related technical fields are included in the patent protection scope of the present invention.

[0144] The present disclosure also provides a computer-readable storage medium storing a computer program that, when executed by a processor, executes the steps of the method for intelligently monitoring the construction progress of a decorative curtain wall building, as described in the above method embodiment. The storage medium can be either volatile or non-volatile, computer-readable.

[0145] In addition, an embodiment of the present disclosure also provides a computer program product, which stores a computer program. When the computer program is run by a processor, it executes the steps of an intelligent monitoring method for the construction progress of a building decorative curtain wall provided in any of the above embodiments of the present disclosure. For details, please refer to the above method embodiments, which will not be repeated here.

[0146] The computer program product may be implemented in hardware, software, or a combination thereof. In one embodiment, the computer program product is implemented as a computer storage medium, which may be a volatile or non-volatile computer-readable storage medium. In another embodiment, the computer program product is implemented as a software product, such as a software development kit (SDK).

[0147] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described equipment and devices can refer to the corresponding processes in the aforementioned method embodiments, and will not be repeated here. In the several embodiments provided in the present disclosure, it should be understood that the disclosed equipment, devices and methods can be implemented in other ways. The device embodiments described above are merely schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some communication interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0148] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0149] In addition, each functional unit in each embodiment of the present disclosure may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0150] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium that is executable by a processor. Based on this understanding, the technical solution of the present disclosure, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, which is stored in a storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present disclosure. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0151] Finally, it should be noted that the above-described embodiments are only specific implementation methods of the present disclosure, which are used to illustrate the technical solutions of the present disclosure, rather than to limit them. The scope of protection of the present disclosure is not limited thereto. Although the present disclosure has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that any person skilled in the art can modify or easily conceive of changes to the technical solutions described in the above-mentioned embodiments within the technical scope disclosed in the present disclosure, or replace some of the technical features therein with equivalents. Such modifications, changes, or replacements do not deviate from the spirit and scope of the technical solutions of the embodiments of the present disclosure, and should be included in the scope of protection of the present disclosure. Therefore, the scope of protection of the present disclosure should be based on the scope of protection of the claims.

Claims

1. A method for intelligently monitoring the construction progress of a building decorative curtain wall, characterized in that: include: Use IoT devices and high-definition cameras to collect environmental parameters, material status, personnel location, and construction images of the construction site to obtain multi-source construction data; Constructing a multidimensional tensor model, and applying a local reversible mapping algorithm in the multidimensional tensor model to extract data association features of the multi-source construction data to generate a standardized feature data set; Using the standardized feature data set, parallel computing is performed through shared memory hierarchical process mapping to evaluate the construction status and output a project progress status evaluation result; Construct a decision diagram based on the project progress status assessment results, compare and analyze the actual progress with the planned progress, identify delay risk points, and generate a construction progress deviation report; Based on the construction progress deviation report, a sample optimal private regression algorithm is used to predict the construction progress trend and generate an optimization plan to form a construction progress forecast report; The construction progress forecast report is displayed through a visual interface, and an early warning is issued according to a preset threshold.

2. The method according to claim 1, characterized in that IoT devices and high-definition cameras collect environmental parameters, material status, personnel location, and construction images of the construction site to obtain multi-source construction data, including: Deploy environmental monitoring sensors, material tracking tags, personnel location devices, and equipment status monitors at construction sites; The collected raw data is compressed, noise filtered, and outlier detected to obtain pre-processed multi-source data streams; The pre-processed multi-source data stream is stored according to time, location and data type to form the multi-source construction data.

3. The method according to claim 1, characterized in that Applying a local reversible mapping algorithm in the multidimensional tensor model to extract data association features of the multi-source construction data and generate a standardized feature data set includes: Map different types of data into a unified tensor space and construct an initial multi-dimensional tensor representation; Applying a local reversible mapping algorithm to the initial multidimensional tensor representation to extract internal correlation features between the multi-source construction data and generate a feature mapping matrix; The feature mapping matrix is ​​used to perform standardization processing on the data and correct outliers, and a standardized feature data set is output.

4. The method according to claim 3, characterized in that Applying a local reversible mapping algorithm to the initial multidimensional tensor representation to extract internal correlation features between the multi-source construction data and generate a feature mapping matrix, including: Based on the initial multidimensional tensor representation, calculating the Euclidean distance between data points and constructing an initial distance matrix; Using the initial distance matrix, a set of nearest neighbor points is selected to establish a local neighbor graph structure; Based on the local neighbor graph structure, calculating the local reconstruction weight of each data point to form a weight matrix; Using the weight matrix, a global covariance matrix is ​​constructed to solve the eigenvalue decomposition problem; Based on the eigenvalue decomposition result, selecting main eigenvectors to obtain a low-dimensional embedding representation; A nonlinear transformation is performed on the low-dimensional embedding representation to generate a feature mapping matrix.

5. The method according to claim 1, wherein By using the standardized feature data set, parallel computing is performed through shared memory hierarchical process mapping to evaluate the construction status and output the project progress status evaluation results, including: Partition data and create indexes to form a data structure that supports parallel access; Build a hierarchical process mapping framework to define the mapping relationship between data processing tasks and computing resources; Use historical construction data to train construction period prediction models, resource consumption models, and quality assessment models, evaluate the current construction status, and output project progress status assessment results.

6. The method according to claim 1, characterized in that Based on the project progress status assessment results, a decision diagram is constructed to compare and analyze the actual progress with the planned progress, identify delay risk points, and generate a construction progress deviation report, including: Construct a decision diagram based on integrated tensor representation and establish a construction progress evaluation indicator system; Matching the preset construction plan progress data with the decision diagram to establish a corresponding relationship matrix between the planned progress and the actual progress; The progress difference is calculated through the corresponding relationship matrix, key delay points and influencing factors are identified, and a construction progress deviation report is generated.

7. The method according to claim 1, characterized in that Based on the construction progress deviation report, a sample optimal private regression algorithm is used to predict the construction progress trend and generate an optimization plan to form a construction progress forecast report, including: Combine historical construction data to construct a sample data space and determine the prediction feature dimensions; Applying a polynomial-time sample-optimal private regression algorithm in the sample data space to train a progress prediction model; The progress prediction model is used to simulate progress changes under different adjustment strategies, select the optimal solution, and generate a construction progress prediction report.

8. The method according to claim 7, characterized in that Applying a polynomial-time sample-optimal private regression algorithm in the sample data space to train a progress prediction model includes: Based on the sample data space, a time window matrix is ​​constructed, and the historical construction data is segmented to obtain a training sample set; Using the training sample set, calculating the similarity weights between samples and constructing a weighted distance matrix; Based on the weighted distance matrix, applying a kernel function to perform feature mapping to obtain a high-dimensional feature space; In the high-dimensional feature space, iteratively optimizing regression parameters by gradient descent method to construct a private regression model; Cross-validating the private regression model, selecting an optimal model parameter combination, and obtaining a validated regression model; The verified regression model is applied to the current construction data to output the progress prediction result.

9. The method according to claim 1, characterized in that After generating the standardized feature dataset, it also includes: Performing dimensionality reduction and information compression on the standardized feature dataset, including: Based on the standardized feature dataset, applying a tensor decomposition algorithm to separate high-dimensional features; Perform principal component analysis on the separated high-dimensional features and select the principal components whose feature contribution rate exceeds the preset threshold; The feature data is reconstructed based on the selected principal components to obtain the key feature set after dimensionality reduction.

10. An intelligent monitoring system for the construction progress of a building decoration curtain wall, characterized in that: include: The data acquisition module is used to collect environmental parameters, material status, personnel location and construction images of the construction site through IoT devices and high-definition cameras to obtain multi-source construction data; A data processing module, configured to construct a multidimensional tensor model for the multi-source construction data and apply a local reversible mapping algorithm to extract data association features to generate a standardized feature data set; A parallel computing module is used to utilize the standardized feature data set to perform parallel computing through shared memory hierarchical process mapping, evaluate the construction status, and output a project progress status evaluation result; A progress analysis module is used to construct a decision diagram based on the project progress status assessment results, compare and analyze the actual progress with the planned progress, identify delay risk points, and generate a construction progress deviation report; A prediction and optimization module, configured to predict the construction progress trend and generate an optimization plan based on the construction progress deviation report using a sample optimal private regression algorithm to form a construction progress prediction report; The display and warning module is used to display the construction progress forecast report through a visual interface and issue warnings based on preset thresholds.

Images