Digital management system and method for whole process of cost

Abstract

The invention relates to the field of engineering project cost management, and discloses a cost whole process digital management system and method, and the system comprises a data collection and processing module which is used for fusing multi-source heterogeneous data and constructing a space-time tensor; the intelligent analysis module is used for extracting a state vector by adopting tensor decomposition and establishing a dynamic system equation; the execution control module is used for generating a control instruction based on an H-infinity robust control strategy and a hypergraph model; and the decision interaction module realizes three-dimensional thermodynamic diagram visualization and closed-loop data updating, and the multi-dimensional spatio-temporal data tensor has a four-order structure. According to the method, the fourth-order space-time tensor model is constructed, H-infinity robust control and three-dimensional thermodynamic diagram visualization are introduced, data fusion, dynamic regulation and control and intelligent decision making in the whole process of the project cost are achieved, and the real-time performance and the cooperation efficiency of the system are improved.

Description

Technical Field

[0001] This invention relates to the field of engineering project cost management technology, and in particular to a digital management system and method for the entire cost process. Background Technology

[0002] With the deep integration of Building Information Modeling (BIM) and Internet of Things (IoT) technologies, modern engineering cost management is gradually transforming from a two-dimensional budgeting model to a multi-dimensional dynamic control model. Engineering cost data exhibits high-dimensionality and strong spatiotemporal correlation characteristics, involving the coupled effects of multiple factors such as schedule planning, resource scheduling, and environmental parameters. Traditional static analysis methods based on relational databases are no longer sufficient to meet the needs of refined management.

[0003] Most current mainstream cost management systems adopt a layered architecture: the bottom layer integrates progress and quantity data through a BIM platform; the middle layer deploys an SQL database to store resource prices and consumption records; and the application layer provides cost statistics and report generation functions. Data updates are performed using timed batch processing, and control strategies trigger alerts based on deviation thresholds matched with preset rule bases. The visualization module typically displays cost trends using a combination of Gantt charts and bar charts.

[0004] While existing technologies have achieved BIM data integration and cost analysis functions in engineering cost management, some shortcomings remain. First, traditional systems use two-dimensional relational databases to manage cost data, which cannot effectively express the coupling relationships between multi-dimensional information such as time, space, and resources, resulting in low efficiency and large errors when processing heterogeneous data fusion. Second, existing technologies generally rely on static rule bases or experience models for cost deviation analysis, lacking dynamic system modeling capabilities and making it difficult to predict the evolution trend of resource consumption during construction. In addition, their control strategies mostly adopt fixed-parameter PID control, which lacks robustness in the face of market fluctuations or construction disturbances, easily causing control instability or sluggish response. Finally, the visualization methods are usually based on two-dimensional charts and have not been deeply integrated with the BIM three-dimensional spatial model, making it difficult for managers to intuitively identify the spatial distribution characteristics of cost deviations, affecting problem location and response efficiency. Summary of the Invention

[0005] The purpose of this invention is to provide a digital management system and method for the entire cost process, which solves the problems of difficulty in multi-source data fusion, insufficient dynamic modeling capabilities, and low efficiency of human-machine collaborative decision-making in the existing technology.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a digital management system for the entire cost estimation process, characterized in that it comprises: The data acquisition and processing module uses a spatiotemporal alignment engine to fuse multi-source heterogeneous data into a multi-dimensional spatiotemporal data tensor and performs missing data completion. The intelligent analysis module receives the data tensor and performs tensor decomposition to extract the dimensionality-reduced state vector to establish the dynamic system state equation; The execution control module solves for a robust control strategy based on the state equations and dynamically reconstructs the service topology through the hypergraph service bus to execute control commands. The decision interaction module maps the execution results to the visualization interface and sends the feedback data back to the data acquisition and processing module to update the data tensor; The multidimensional spatiotemporal data tensor has a fourth-order structure. Where T represents the time dimension; S represents the engineering spatial grid division; R represents the resource type set; and E represents the environmental impact factors.

[0007] This invention also provides a method for digital management of the entire cost process, comprising the following steps: Construct a fourth-order data tensor through spatiotemporal alignment and fill in missing values; Perform tensor decomposition to extract state features; Establish a dynamic system model containing uncertainties; Comprehensive robust H∞ control strategy; Dynamically reconstruct the hypergraph service topology; Visual mapping and feedback updates.

[0008] In summary, the present invention has at least one of the following beneficial technical effects: 1. This invention solves the data silo problem in traditional cost management systems by constructing a spatiotemporal fourth-order tensor model to fuse heterogeneous information such as BIM model data, IoT sensor data, and supply chain data under a unified spatiotemporal coordinate system. It innovatively introduces a tensor completion algorithm with truncated nuclear norm constraints to effectively repair missing data caused by sensor failures or communication interruptions, thereby improving data quality.

[0009] 2. This invention, based on improved Tucker decomposition to extract core state vectors and combined with Jacobian matrices to construct state transition equations, overcomes the limitations of traditional static cost analysis. This modeling method can capture the spatiotemporal evolution of resource consumption, providing a dynamic evolution model for predicting the risk of project cost overruns.

[0010] 3. This invention designs a robust controller based on the H∞ performance index and obtains the optimal control strategy by solving linear matrix inequalities. This control method has a strong inhibitory effect on disturbances such as market price fluctuations and construction delays, ensuring that the actual cost trajectory closely tracks the budget plan.

[0011] 4. This invention innovatively employs a hypergraph model to represent the dynamic relationships between service nodes, adaptively adjusting the service topology through a relationship threshold. Compared to a fixed service chain structure, this mechanism can quickly respond to changes in resource supply and demand, improving supply chain collaboration efficiency.

[0012] 5. This invention develops a BIM-based augmented reality visualization system that maps abstract control quantities into spatial heat maps and overlays them onto a 3D model. This human-computer interface supports multi-dimensional data drill-down analysis, helping managers quickly pinpoint the sources of cost deviations and improve decision-making response speed. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the system structure of the present invention; Figure 2 This is a schematic diagram of the module architecture of the data acquisition and processing module of the present invention; Figure 3 This is a schematic diagram of the module architecture of the intelligent analysis module of the present invention; Figure 4 This is a schematic diagram of the module architecture of the execution control module of the present invention; Figure 5 This is a schematic diagram of the module architecture of the decision-making interaction module of the present invention; Figure 6 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0014] The following is in conjunction with the appendix Figure 1 - Appendix Figure 6 The present invention will be further described in detail below.

[0015] This invention provides a digital management system for the entire cost process. By constructing a fourth-order spatiotemporal tensor model, introducing H∞ robust control and three-dimensional heat map visualization, it realizes data fusion, dynamic control and intelligent decision-making throughout the entire cost process.

[0016] like Figure 1-5 As shown, this digital management system for the entire cost process may include: The data acquisition and processing module uses a spatiotemporal alignment engine to fuse multi-source heterogeneous data into a multi-dimensional spatiotemporal data tensor and performs missing data completion. In this embodiment, the data acquisition and processing module achieves the fusion and governance of multi-source heterogeneous data by constructing a multi-dimensional spatiotemporal data tensor. This module includes a spatiotemporal alignment engine and a tensor builder, and the specific implementation process is as follows: The spatiotemporal alignment engine in this embodiment establishes a unified spatiotemporal coordinate system to perform spatiotemporal registration of heterogeneous data from BIM models, IoT sensors, and supply chain systems. The time dimension is defined as a discretized sequence T = {t} i |t i =t0+iΔt,i=0,1,...,N-1}, where Δt is the preset sampling period; t0 is the project start timestamp.

[0017] Spatial dimension S = {sj The BIM model components are divided into grids, with each grid cell s. j =(x j ,y j ,z j (This refers to the spatial location of a specific part of the project.)

[0018] In this embodiment, the resource dimension R = {r k It includes three basic resource types: labor, materials, and machinery, with each type further subdivided into multiple subcategories. The environmental dimension E = {e...} l It covers external influencing factors such as temperature, humidity, and market price index, with the market price index obtaining updated data through a real-time API interface connected to the supply chain system.

[0019] The tensor builder in this embodiment executes fourth-order tensors. Where T represents the time dimension; S represents the engineering spatial grid division; R represents the resource type set; and E represents the environmental impact factors. For missing locations where no data was collected, the system marks them as missing values ​​and stores them in the index set Ω. c ={(i,j,k,l)|x ijkl Missing}.

[0020] The data completion unit in this embodiment uses an optimization method that truncates the nuclear norm constraint to handle missing data, and solves the following objective function: Where Ω represents the effective observation index set; ||·|| TNN The truncation nuclear norm is λ; λ > 0 is the regularization coefficient. This represents the actual observation at position (i, j, k, l); x ijkl Indicates time t i Space s j Resources k Environment e l Cost observations under the given conditions.

[0021] The spatiotemporal alignment engine in this embodiment has a dynamic calibration mechanism. When a sensor clock deviation is detected to exceed a threshold, it will trigger a dynamic calibration mechanism. t When the BIM model version is updated, the following calibration process will be automatically triggered: Time-axis resampling: interpolating historical data to generate a new time series T. ′ Data point spatial grid reconstruction: re-dividing spatial units S based on the updated BIM model ′ And migrate the original data through nearest neighbor interpolation. Resource type mapping: Establish a comparison table between old and new resource codes to maintain data consistency.

[0022] The intelligent analysis module receives data tensors and performs tensor decomposition to extract dimensionality-reduced state vectors to establish dynamic system state equations. In this embodiment, the intelligent analysis module achieves in-depth feature extraction and state evolution analysis of cost data through tensor decomposition and dynamic system modeling. This module includes a tensor decomposition core and a state-space converter, and the specific implementation process is as follows: The tensor decomposition core of this embodiment uses an improved Tucker decomposition method to process the fourth-order spatiotemporal data tensor XX, and its decomposition form is as follows: in, For the core tensor; U T U is the time factor matrix; S U is the spatial factor matrix; R For resource factor matrix; U E This is an environmental factor matrix; × n This is a tensor modulus product operation. The decomposition process is optimized iteratively using alternating least squares, fixing three factor matrices and updating the remaining matrices in each iteration until the reconstruction error converges.

[0023] In this embodiment, the state-space converter extracts dynamic feature vectors from the decomposition results, specifically for the core tensor quantum block corresponding to each time slice k. Perform modal product operation: in, The core tensor quantum block corresponding to time slice k; It is a spatial factor matrix; Let be the resource factor matrix; vec(·) denotes the tensor vectorization operation; x(k) is the system state vector at time k. This process effectively integrates the characteristic information of spatial resource allocation and resource consumption patterns.

[0024] The dynamic system modeling in this embodiment is based on the extracted state sequence x(k) to establish a discrete-time state equation: x(k+1)=Ax(k)+Bu(k)+Γw(k); in, This is the state transition matrix; The input matrix is ​​used for control, where m is the dimension of the control variables; Let p be the interference input matrix, and p be the dimension of the interference terms; To control the input vector; The state transition matrix A is a bounded disturbance vector; its construction satisfies the following conditions: in, Let f be the Jacobian matrix of function f at x(k); x(k) is the state vector at discrete time k; k is the time index.

[0025] The discretization process in this embodiment employs the Tustin bilinear transformation method to convert the continuous-time system into a discrete form. For a given continuous-time system matrix A... c Its discretized form is: Where A is the state transition matrix of the discrete-time system; I is the identity matrix; Δt is the discretization time step; A c Let be the state matrix of the continuous-time system.

[0026] This transformation preserves system stability and frequency domain characteristics, avoiding numerical divergence problems that may be caused by the explicit Euler method.

[0027] The robustness guarantee mechanism in this embodiment is reflected in the boundedness constraint on the disturbance term w(k), setting ‖w(k)‖2≤ρ, where ρ is an upper bound value determined based on historical data statistics. This constraint condition forms a technical closed loop with the subsequent H∞ optimization design of the control module.

[0028] The execution control module solves robust control strategies based on state equations and dynamically reconstructs service topology through the hypergraph service bus to execute control commands. In this embodiment, the execution control module optimizes and regulates the cost estimation process through robust control strategies and dynamic service coordination. This module includes a control law solver and a hypergraph service bus, and the specific implementation process is as follows: The robust control strategy in this embodiment is designed based on the H∞ performance index, and the state feedback gain matrix is ​​obtained by solving linear matrix inequalities. The matrix inequality form is as follows: Where P>0 is a symmetric positive definite matrix, γ>0 is the H∞ performance index, and I is the identity matrix; This is the state transition matrix; To control the input matrix; It is a symmetric positive definite matrix; For H∞ performance indicators; The matrix is ​​an identity matrix with dimensions consistent with the control input space. The solution process employs a semidefinite programming algorithm, using the YALMIP toolbox to resolve matrix constraints and then calling the MOSEK solver for numerical optimization.

[0029] The hypergraph service bus in this embodiment defines dynamic hyperedge generation rules, and its hyperedge set construct is as follows: ε={e j |e j ={vi |φ(v i ,v j )>θ}}; Where φ(·) is the cost correlation function; θ is the dynamically adjusted connection threshold; φ(v i ,v j ) is the service node correlation function.

[0030] In this embodiment, the service topology reconstruction mechanism responds to control command u(k) and performs the following operations: Analyze the control vector u(k) and extract each spatial unit s. j Resource regulation amount u jk According to u jk Value adjustment of superedge weight w e The updated formula is: Where J is the cost bias function, and β is the learning rate parameter; w e Let be the weight value of the hyperedge e; For weight gradients; The historical weight values ​​of the superedge e before the update; The weight value of the updated hyperedge e; Remove hyperedges with a correlation degree lower than θ(t) and add new ones that satisfy φ(v). i ,v j Hyperedge of θ(t) In this embodiment, the interference suppression term Γw(k) is implemented through a feedforward compensation mechanism, specifically by constructing the interference observer: in, Let Γ be the pseudo-inverse matrix; Γ is the disturbance estimation vector at time k; x(k+1) is the actual state vector at time k+1; Γ is the disturbance input matrix; (x(k+1)-Ax(k)-Bu(k)) is the state residual vector, reflecting the combined effects of unmodeled dynamics and external disturbances.

[0031] The observation results are used to update the upper bound estimate of the disturbance online, thereby enhancing the robustness of the system.

[0032] In this embodiment, the control command distribution adopts a publish-subscribe pattern, using a message middleware to decompose u(k) into operation commands for each execution terminal. Preferably, the command encoding format is JSON-LD, which includes three information dimensions: spatial location, resource type, and control quantity.

[0033] The service bus in this embodiment has a fault tolerance mechanism. When node v is detected, i When a response times out or the execution deviation exceeds a threshold, the following actions are automatically triggered: Temporarily mask v in the hyperedge set i node; Start standby node v′ i And recalculate the correlation. Control commands are redistributed based on the latest topology.

[0034] The decision interaction module maps the execution results to the visualization interface and sends the feedback data back to the data acquisition and processing module to update the data tensor; In this embodiment, the decision-making interaction module achieves human-machine collaborative decision-making through visual mapping and feedback data closed-loop. This module includes a 3D rendering engine and a data feedback controller, and the specific implementation process is as follows: In this embodiment, the visualization interface maps the control command u(k) to a spatial heatmap, and its rendering formula is as follows: Among them, u jk In space s j For resource r k Control quantity; H(s) j ) represents the spatial unit s j The overall control intensity. The heatmap is encoded using the HSL color space, with hue values ​​h. j With the regulation intensity H(s) j A linear relationship exists: Among them, h j For spatial unit s j The corresponding HSL hue value; H min With H max These represent the minimum and maximum values ​​of the control intensity of all spatial units within the current time slice.

[0035] In this embodiment, the BIM model overlay uses a ray casting algorithm to map the heatmap texture onto the surface of the 3D component. For each spatial unit s j Calculate the coordinates of its bounding box vertices {v m}, and the thermal value H(s) is obtained by UV expansion. j The coordinates are mapped to the texture coordinates of the corresponding region. Preferably, real-time rendering on the browser side is implemented using the WebGL 2.0 API.

[0036] The feedback data update mechanism in this embodiment responds to the execution result and dynamically corrects the fourth-order tensor XX. The correction formula is: x(k,:,:,c 实际 )=x(k,:,:,c 计划 )⊙M Δ +Δx 偏差 ; Among them, M Δ The mask matrix for the difference between planned and actual values; Δx 偏差 This refers to the actual execution deviation data collected by sensors; x(k,:,:,c 实际 x(k,:,:,c) represents the tensor quantum block corresponding to the actual environmental factors in the k-th time slice; 计划 () represents the original data sub-block corresponding to the planned environmental factors for the k-th time slice. Correction operations are completed before the start of the next control cycle to ensure data timeliness.

[0037] The user interface in this embodiment features a multi-dimensional filter that supports data drill-down based on time dimension T, spatial dimension S, and resource dimension R. The filter condition combination Φ = (T... q ,S q ,R q The corresponding formula for extracting a subset of data is: x Φ =x(T) q ,S q ,R q ,:); Where, x Φ T represents the subset of data that meets the filtering condition Φ; q Time-based query criteria; S q The spatial dimension query criteria are the set of selected component IDs in the BIM model; R q The query criteria are for resource dimensions, and are a subset of resource types specified by the user.

[0038] The subset is visualized using a parallel coordinate system to show the correlation between each environmental factor (EE) and cost indicators.

[0039] The early warning mechanism in this embodiment is based on the control deviation Δu = u 实际 -u 指令 The instruction sets a threshold condition, and when ||Δu|2>∈ is satisfied, the following operation is triggered: Highlight spatial units with excessive deviations in the 3D model; Automatically generate deviation analysis reports, including resource type distribution and time evolution trends; Pushing early warning information to the terminal devices of relevant responsible parties; Where Δu is the control command deviation vector; ∈ is the deviation threshold.

[0040] In this embodiment, the feedback loop is connected to the acquisition module via a data pipeline, forming an iterative optimization process of "control command → execution result → data update → re-control". Preferably, the Apache Kafka message queue is used to achieve high-throughput real-time data synchronization.

[0041] The method for digital management of the entire cost process described below can be referred to in correspondence with the digital management system for the entire cost process described above.

[0042] Please see the appendix Figure 6 The present invention also provides a method for digital management of the entire cost process, comprising the following steps: Construct a fourth-order data tensor through spatiotemporal alignment and fill in missing values; Perform tensor decomposition to extract state features; Establish a dynamic system model containing uncertainties; Comprehensive robust H∞ control strategy; Dynamically reconstruct the hypergraph service topology; Visual mapping and feedback updates.

[0043] The method in this embodiment can be used to execute the above system embodiment, and its principle and technical effect are similar, so it will not be described again here.

[0044] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A digital management system for the entire cost estimation process, characterized in that, include: The data acquisition and processing module uses a spatiotemporal alignment engine to fuse multi-source heterogeneous data into a multi-dimensional spatiotemporal data tensor and performs missing data completion. The intelligent analysis module receives the data tensor and performs tensor decomposition to extract the dimensionality-reduced state vector to establish the dynamic system state equation; The execution control module solves for a robust control strategy based on the state equations and dynamically reconstructs the service topology through the hypergraph service bus to execute control commands. The decision interaction module maps the execution results to the visualization interface and sends the feedback data back to the data acquisition and processing module to update the data tensor.

2. The digital management system for the entire cost process according to claim 1, characterized in that, The data acquisition and processing module includes: The spatiotemporal alignment engine performs spatiotemporal registration on heterogeneous data from BIM models, sensor networks, and supply chain systems to construct a fourth-order tensor. Where T represents the time dimension; S represents the engineering spatial grid division; R represents the resource type set; and E represents the environmental impact factors. The data completion unit performs tensor completion operations on missing data.

3. The digital management system for the entire cost process according to claim 1, characterized in that, The data completion unit achieves missing value filling by solving the following optimization problem: Where Ω represents the effective observation index set; ||·|| TNN The truncation nuclear norm is λ; λ > 0 is the regularization coefficient. This represents the actual observation at position (i, j, k, l); x ijkl Indicates time t i Space s j Resources k Environment e l Cost observations under the given conditions.

4. The digital management system for the entire cost process according to claim 1, characterized in that, The intelligent analysis module employs an improved Tucker decomposition method when performing tensor decomposition: in, For the core tensor; U T U is the time factor matrix; S U is the spatial factor matrix; R For resource factor matrix; U E This is an environmental factor matrix; × n For tensor modulo product operation; Factor matrix U T U S U R U E Satisfying column orthogonality constraints And d t <<T,d s << S; where T is the original size of the time dimension; S is the original size of the spatial dimension; d t The time factor matrix U T The number of columns; d s For the space factor matrix U S The number of columns; This is a time factor matrix; It is a spatial factor matrix; Let d be the resource factor matrix. r These are preset resource feature dimensions; For the environmental factor matrix, d e These are preset environmental feature dimensions.

5. The digital management system for the entire cost process according to claim 1, characterized in that, The process by which the intelligent analysis module establishes the dynamic system state equation includes: Extract the state vector from the decomposed core tensor: in, The core tensor quantum block corresponding to time slice k; It is a spatial factor matrix; Let be the resource factor matrix; vec(·) denotes the tensor vectorization operation; x(k) is the system state vector at time k; Construct the discrete-time state equation: x(k+1)=Ax(k)+Bu(k)+Γw(k); in, This is the state transition matrix; The input matrix is ​​used for control, where m is the dimension of the control variables; Let p be the interference input matrix, and p be the dimension of the interference terms; To control the input vector; The state transition matrix A is a bounded disturbance vector; its construction satisfies the following conditions: in, Let f be the Jacobian matrix of function f at x(k); x(k) is the state vector at discrete time k; k is the time index.

6. The digital management system for the entire cost process according to claim 1, characterized in that, The robust control strategy obtains the gain matrix K by solving the following linear matrix inequality: Where P>0 is a symmetric positive definite matrix, γ>0 is the H∞ performance index, and I is the identity matrix; This is the state transition matrix; To control the input matrix; It is a symmetric positive definite matrix; For H∞ performance indicators; It is an identity matrix with dimensions consistent with the control input space.

7. The digital management system for the entire cost process according to claim 1, characterized in that, The hypergraph service bus defines a set of hyperedges: e={e j |e j ={v i |φ(v i ,v j )>θ}}; Where φ(·) is the cost correlation function; θ is the dynamically adjusted connection threshold; φ(v i ,v j ) is the service node correlation function.

8. The digital management system for the entire cost process according to claim 1, characterized in that, The visual interface is implemented as follows: Map the control command u(k) to a three-dimensional heat map H(s). j )=∑r k u jk And superimposed on the spatial grid S of the BIM model, where u jk In space s j For resource r k Control quantity; H(s) j ) represents the spatial unit s j The overall control intensity.

9. The digital management system for the entire cost process according to claim 1, characterized in that, The feedback data update process includes: Modify the data tensor based on the execution result. corresponding slice Where k is the current time index.

10. A method for digital management of the entire cost process, characterized in that, Using the full-process digital cost management system according to any one of claims 1-9 includes the following steps: Construct a fourth-order data tensor through spatiotemporal alignment and fill in missing values; Perform tensor decomposition to extract state features; Establish a dynamic system model containing uncertainties; Comprehensive robust H∞ control strategy; Dynamically reconstruct the hypergraph service topology; Visual mapping and feedback updates.

Images