Building industry resource matching and performance warning system based on business data fusion

Abstract

This invention relates to the field of engineering management informatization and data processing technology, and discloses a construction industry resource matching and performance early warning system based on business data fusion. The system includes a multi-source heterogeneous data mapping module that constructs a resource-time slot matrix and transforms financial, physical, and information flow data into standardized feature vectors; a dynamic performance status entropy calculation module that calculates performance status entropy to quantify risk; a time slot adaptive modulation module that dynamically controls time slot granularity splitting or merging and adjusts data sampling frequency based on entropy gradient; a resource status latch control module that generates latch status bits that determine resource visibility based on a multi-dimensional confidence consensus protocol; and a topology path matching and risk blocking module that plans the minimum entropy increase path in the activated resource pool and directly modifies the latch bits based on risk flux propagation prediction to block risk propagation. This invention, based on entropy gradient adaptive modulation and status latching mechanisms, achieves unified quantification of resource risks, dynamic scaling of monitoring granularity, and proactive blocking of cascading risks.

Description

Technical Field

[0001] This invention relates to the field of engineering management informatization and data processing technology, specifically a construction industry resource matching and performance early warning system based on business data fusion. Background Technology

[0002] With the deepening of digital transformation in the construction industry, the supply chain management of large-scale engineering projects has become highly complex and dynamic, involving the coordinated scheduling of massive numbers of labor teams, machinery and equipment, and material suppliers. Against this backdrop, how to leverage data-driven methods to achieve precise matching of supply chain resources and provide real-time early warnings of potential performance risks has become crucial for ensuring on-time project delivery, cost control, and construction safety. It is also a core issue that urgently needs to be addressed in the fields of smart construction sites and engineering management informatization.

[0003] Existing construction resource management and scheduling solutions primarily rely on a combination of Enterprise Resource Planning (ERP) systems, project management software, and Building Information Modeling (BIM) technology. Common technical approaches typically include: using ERP systems to record financial transactions and contract information; using schedule management software to create construction schedule network diagrams based on the Critical Path Method (CPM) or Program Review and Evaluation Technique (PRE); and combining this with on-site Internet of Things (IoT) devices (such as turnstiles and locators) to collect discrete attendance or location data. In the resource matching phase, existing systems often employ weighted scoring models based on historical data to tier and screen suppliers, and track project progress and update resource scheduling at fixed time intervals (such as daily or weekly) according to pre-defined schedule constraints.

[0004] While existing technologies have achieved digitalization and visualization of project management to some extent, some shortcomings remain: Because the construction supply chain involves multi-dimensional and heterogeneous data such as cash flow, physical flow, and information flow, existing technologies lack unified physical metrics to deeply integrate these disparate data. This makes it difficult to quantify and identify hidden risks such as broken cash flow through simple physical progress monitoring. Furthermore, existing systems often use fixed time granularity for monitoring and scheduling. This rigid time-domain sampling mode cannot adapt to the non-linear and abrupt changes in project risks, often missing key precursors during high-risk periods due to sparse sampling, or wasting computing power during stable periods due to oversampling. In addition, traditional weighted average resource evaluation methods are prone to high scores masking low scores, allowing resources with fatal flaws in a key dimension (such as financial status) to still enter the system through comprehensive scoring. Moreover, the lack of a dynamic transmission and prediction mechanism for risk cascading effects means the system can only passively correct progress when facing local risks, making it difficult to proactively block and defend against them. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a construction industry resource matching and performance early warning system based on business data fusion. This system solves the problems of difficulty in quantifying multi-source heterogeneous data fusion in existing construction resource scheduling, the inability of fixed time-domain granularity to adapt to sudden risk changes, and the lack of proactive risk prevention mechanisms for cascading effects.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a construction industry resource matching and performance early warning system based on business data fusion, the system comprising: The multi-source heterogeneous data mapping module is configured to build a digital basic model of construction supply chain resources. It discretizes labor, equipment and material resources in the time dimension to construct a resource-time slot matrix, and converts the capital flow data, physical flow data and information flow data collected from external interfaces into standardized feature vectors respectively. The dynamic performance status entropy calculation module is communicatively connected to the multi-source heterogeneous data mapping module and is configured to receive the feature vector and calculate the performance status entropy of each resource time slot. The performance status entropy is used to quantify the system disorder and performance risk value of a specific resource within a specific time window. The time slot adaptive modulation module is connected to the dynamic performance state entropy calculation module and the external data acquisition interface respectively. It is configured to monitor the entropy gradient of the performance state entropy in real time, and dynamically control the physical granularity of the resource time slot to perform splitting or merging operations according to the change amplitude of the entropy gradient, and synchronously send instructions to the external data acquisition interface to adjust the sampling frequency of the underlying data. The resource status latch control module is configured at the front end of the topology path matching and risk blocking module. It is configured to execute a confidence consensus protocol based on multidimensional data to generate binary latch status bits. The latch status bits determine the visibility and availability of resource slots in the system. The topology path matching and risk blocking module is configured to construct a directed acyclic graph of project tasks. In the resource pool that has been filtered by the resource status latch control module and whose latch status bit is active, it finds the resource configuration path with the minimum increase in total system entropy and is configured to calculate the network propagation of risk flux. When it is predicted that a downstream node is facing high risk, it directly modifies the latch status bit of the associated resource to cut off the risk propagation path.

[0007] Preferably, the multi-source heterogeneous data mapping module includes a heterogeneous data vectorization processing unit, which is configured as follows: Payment and credit records associated with specific resources are extracted from the financial system to generate a fund flow vector. The features of the fund flow vector aggregation include the normalized average bill of exchange term, the normalized historical cumulative payment default frequency, and the inverse of the account liquidity ratio corresponding to the current time slot. Physical flow vectors are generated based on real-time data collected by on-site sensor networks. The features of the aggregated physical flow vectors include the average load rate of the equipment in the current time slot, the ratio of actual working hours to standard working hours, and the normalized Euclidean distance between the actual geographic coordinates of the resources and the work area specified by the model. The engineering document log is parsed to generate an information flow vector. The features of the aggregated information flow vector include the number of quality and safety rectification notices associated with the current time slot and the average lag time of the engineering acceptance document signing process.

[0008] Preferably, the dynamic performance state entropy calculation module calculates the performance state entropy using a hybrid entropy model, and the calculation process includes: Using a Bayesian update algorithm based on time decay factor, the posterior probability distribution of historical performance status events is calculated, and the Shannon entropy component is calculated accordingly. A diagonal weight matrix for the capital flow dimension is introduced to perform a linear transformation on the capital flow vector, and the Euclidean norm of the transformed vector is calculated as the capital entropy component. A comprehensive deviation scalar is calculated and mapped using a nonlinear exponential function as a physical deviation entropy component. The comprehensive deviation scalar is a weighted composite of the position deviation in the physical flow vector and the number of rectification orders in the information flow vector. The Shannon entropy component, the capital entropy component, and the physical deviation entropy component are linearly weighted and aggregated to obtain the performance status entropy.

[0009] Preferably, the time slot adaptive modulation module includes an entropy gradient calculation subroutine and time slot splitting execution logic; The entropy gradient calculation subroutine is configured to read the performance status entropy of the current time slot and the previous time slot, and calculate the entropy gradient through discrete difference operation; The time slot splitting execution logic is configured such that when the absolute value of the entropy gradient exceeds a preset splitting threshold, the single time slot object with the original time span as the initial standard time granularity is decomposed into multiple consecutive micro-time slot units, and a synchronous acquisition command is sent to the multi-source heterogeneous data mapping module through the reverse control link to forcibly increase the data acquisition frequency for the specific resource in order to match the accuracy of the split micro-time slots.

[0010] Preferably, the time slot adaptive modulation module further includes time slot merging callback logic; The time slot merging callback logic is configured to maintain a sliding time window. When the absolute value of the average entropy gradient of multiple consecutive periods within the sliding time window is lower than the preset merging threshold, the adjacent micro time slot units are re-aggregated into time slot objects with standard time granularity, and instructions are sent synchronously to reduce the acquisition frequency of the underlying data. The set value of the merging threshold is less than the set value of the splitting threshold to form a hysteresis comparison interval.

[0011] Preferably, the resource state latch control module is configured to execute the following confidence consensus protocol to determine the latch state bits: The capital flow vector, the physical flow vector, and the information flow vector are mapped to capital confidence scores, physical confidence scores, and information confidence scores, respectively. Preset minimum survival thresholds for the financial, physical, and informational dimensions respectively; Using a logical AND operation, the latched state bit is marked as active only when the financial confidence score, the physical confidence score, and the information confidence score are all greater than or equal to their respective minimum survival thresholds; otherwise, the latched state bit is marked as frozen, and a masking code is applied to the resource time slots in the frozen state at the system logic layer.

[0012] Preferably, the topology path matching and risk blocking module is configured to run a path optimization algorithm, the objective function of which is configured to find a resource allocation combination such that the sum of the performance state entropy of all selected resource slots in the candidate resource allocation scheme set reaches the minimum value.

[0013] Preferably, the topology path matching and risk blocking module includes a risk throughput calculation engine; The risk flux calculation engine is configured to compare the actual running status and planned status of task nodes in real time. When it is detected that the performance status entropy of a certain node in actual execution exceeds the baseline predicted entropy value of the node in the initial matching stage, the difference between the two is calculated as the risk flux. The risk flux represents the amount of energy that local disorder spreads to the system.

[0014] Preferably, the topology path matching and risk blocking module further includes a cascade effect blocking controller, which is configured as follows: Calculate the impact of the risk flux of the upstream node on the future state of the downstream associated node to update the predicted entropy value of the downstream node; The updated prediction entropy value of the downstream node is obtained by superimposing the initial prediction entropy value of the downstream node with the propagation influence value of all preceding dependent nodes; The calculation of the transmission impact value is based on the risk flux of the upstream dependent node, the correlation impact coefficient from the upstream node to the downstream node, and the natural exponential decay term based on the time damping factor. The time variable of the natural exponential decay term is the difference between the planned start time of the downstream node and the planned end time of the upstream node.

[0015] Preferably, the cascade effect blocking controller is further configured as follows: When the predicted entropy gradient value calculated based on the updated predicted entropy value exceeds the splitting threshold set by the time slot adaptive modulation module, a blocking command is generated and fed back to the resource status latching control module. In response to the blocking command, the resource status latch control module forcibly overwrites the latch status bit of the originally matched resource of the downstream node to the locked state, and triggers a rematch interrupt signal to start the alternative resource optimization procedure.

[0016] This invention provides a construction industry resource matching and performance early warning system based on business data fusion. It has the following beneficial effects: 1. This invention constructs a multi-source heterogeneous data mapping model and a dynamic performance state entropy calculation mechanism to uniformly map the originally discrete and heterogeneous capital flow, physical flow and information flow data in the construction supply chain into standardized feature vectors. It also uses information entropy theory to quantify the system disorder of resource performance, thereby solving the technical pain point of difficulty in integrating and quantifying multi-dimensional data in traditional construction management and providing a unified and objective risk measurement benchmark for different types of resources.

[0017] 2. This invention monitors entropy gradient changes in real time through a time-slot adaptive modulation module, establishing an elastic scaling mechanism for resource spatiotemporal granularity. It can automatically perform time-slot splitting and synchronously increase the underlying data acquisition frequency during periods of sudden risk changes, while performing time-slot merging during stable periods to release computing resources. This reduces the system's computing load and transmission bandwidth pressure without missing high-frequency risk signals, achieving an adaptive balance between monitoring accuracy and system performance.

[0018] 3. This invention executes a confidence consensus protocol based on multidimensional data through a resource status latch control module. It adopts strict logic and operation requirements that the confidence scores of all dimensions simultaneously meet the minimum survival threshold, thereby generating binary latch status bits that control resource visibility. This prevents the situation in the traditional weighted evaluation system where a high score in one dimension masks fatal risks in other dimensions (such as funding shortages), and constructs a resource access barrier based on strict data trust.

[0019] 4. This invention calculates the network transmission of risk flux through topology path matching and risk blocking modules, uses a cascading effect model with time damping factor to deduce the dynamic impact of upstream risks on downstream nodes, and triggers a pre-blocking mechanism to directly overwrite the latched state bit when a systemic collapse risk is predicted, thereby realizing the transformation from traditional local static optimization to global dynamic defense, and can proactively cut off the cascading propagation path of risks in the supply chain network. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the structure of the present invention; Figure 2This is a schematic diagram of the logical architecture of the multi-source heterogeneous data mapping module of the present invention; Figure 3 This is a schematic diagram of the data flow and logical architecture of the dynamic performance state entropy calculation module of the present invention; Figure 4 This is a schematic diagram of the control logic and data interaction of the time slot adaptive modulation module of the present invention. Detailed Implementation

[0021] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Please see the appendix Figure 1 , Figure 1 This is a schematic diagram of a building resource matching and early warning system based on entropy gradient-based time-slot adaptive modulation and state latching, according to an embodiment of the present invention. The present invention provides a building resource matching and early warning system based on entropy gradient-based time-slot adaptive modulation and state latching, comprising: a multi-source heterogeneous data mapping module, a dynamic performance state entropy calculation module, a time-slot adaptive modulation module, a resource state latching control module, and a topology path matching and risk blocking module.

[0023] The multi-source heterogeneous data mapping module is used to construct a digital foundation model of construction supply chain resources. This module is configured to discretize labor, equipment, and material resources along the time dimension to construct a resource-time slot matrix, and to convert the cash flow data, physical flow data, and information flow data collected from external interfaces into standardized feature vectors, thereby providing a unified data input foundation for the system.

[0024] The dynamic performance status entropy calculation module communicates with the multi-source heterogeneous data mapping module. The dynamic performance status entropy calculation module is configured to receive feature vectors and, based on information entropy theory combined with a Bayesian probability update model, cash flow weighted calculation, and sensitivity response logic for physical schedule deviations, calculates the performance status entropy for each resource time slot. This performance status entropy is used to quantify the system disorder and performance risk value of a specific resource within a specific time window, unifying multi-dimensional heterogeneous business data into a single physical metric.

[0025] The time-slot adaptive modulation module is connected to both the dynamic performance status entropy calculation module and the external data acquisition interface. The time-slot adaptive modulation module is configured to monitor the changing gradient of the performance status entropy in real time. Based on the magnitude of the entropy gradient change, the module dynamically controls the physical granularity of resource time slots to perform splitting or merging operations, and simultaneously sends instructions to the external data acquisition interface to adjust the sampling frequency of the underlying data. This enables refined frequency domain monitoring for high-risk areas and release of computing power for low-risk areas.

[0026] The resource state latch control module is configured in front of the topology path matching and risk blocking module. As an access controller at the physical layer, it is configured to execute a confidence consensus protocol based on multidimensional data. This protocol requires confidence scores in the financial, physical, and information dimensions to simultaneously meet a minimum survival threshold, thereby generating a binary latch state bit. This latch state bit determines the visibility and availability of resource slots in the system, ensuring that unreliable resources that fail multidimensional verification are physically blocked and cannot enter the subsequent matching process.

[0027] The topology path matching and risk blocking module is used to construct a directed acyclic graph of project tasks. Configured to find the resource allocation path with the minimum increase in total system entropy from the resource pool filtered by the resource state latch control module, while meeting project schedule and cost constraints. Furthermore, this module is also configured to calculate the network propagation of risk flux and trigger a pre-blocking mechanism when a downstream node is predicted to face high risk, directly modifying the latch state bits of the associated resources to cut off the risk propagation path.

[0028] In this invention, a multi-source heterogeneous data mapping module constructs the basic data structure, a dynamic performance state entropy calculation module quantifies risk indicators, a time slot adaptive modulation module dynamically adjusts the system's time resolution based on the risk change rate, a resource state latch control module physically locks unreliable resources based on multi-dimensional data confidence, and finally, a topology path matching and risk blocking module completes optimal path planning and dynamic defense, forming a data-driven adaptive closed-loop control system.

[0029] Please see the appendix Figure 2 , Figure 2 This is a logical architecture diagram of a multi-source heterogeneous data mapping module according to an embodiment of the present invention. The multi-source heterogeneous data mapping module is configured to perform resource discretization and multi-dimensional data feature extraction tasks, and its internal logical units include a resource matrix initialization unit, a multi-protocol data acquisition interface unit, and a heterogeneous data vectorization processing unit.

[0030] The resource matrix initialization unit is responsible for establishing a digital spatiotemporal mapping model of construction supply chain resources. This unit traverses the set of resource pools to be scheduled, identifies the entity attributes of labor teams, machinery and equipment, and material suppliers, and discretizes them on the time axis based on a preset initial standard time granularity to construct a resource-time slot matrix. The resource-time slot matrix is ​​defined as a two-dimensional data structure, where the row dimension corresponds to a unique resource identifier, and the column dimension corresponds to a continuous time window.

[0031] Each element in the matrix is ​​defined as a time slot unit, which not only stores the resource availability status within that time window but also serves as a container for subsequent feature vectors. The initial standard time granularity is set based on the smallest control unit of the project construction plan, typically set to the duration of a work shift or a natural day.

[0032] The multi-protocol data acquisition interface unit works in conjunction with the resource matrix initialization unit to acquire raw business data from external systems through predefined ETL (Extract, Transform, Load) rules. This unit is configured with protocol converters adapted to different data sources, including an SQL data interface for connecting to enterprise resource planning systems, an MQTT protocol interface for connecting to field IoT devices, and a RESTful API interface for connecting to engineering project management systems.

[0033] The heterogeneous data vectorization processing unit is used to map the collected raw data into standardized mathematical vectors, specifically including three parallel processing processes: capital flow vector construction, physical flow vector construction, and information flow vector construction.

[0034] In the process of constructing the cash flow vector, the heterogeneous data vectorization processing unit extracts payment and credit records associated with specific resources from the financial system to generate the cash flow vector. This vector aggregates multiple dimensions reflecting the health of funds, including the average payment period of accepted bills, the frequency of historical payment defaults, and the current account liquidity ratio. The formula for constructing the cash flow vector is as follows: ; In the formula, Represents the cash flow vector; This represents the average payment period for accepted bills of exchange after normalization. This represents the normalized historical cumulative frequency of payment defaults; This represents the reciprocal of the account liquidity ratio corresponding to the current time slot.

[0035] In the physical flow vector construction process, the heterogeneous data vectorization processing unit generates a physical flow vector based on real-time data collected from the field sensor network. This vector aims to reflect the deviation between the actual operating status of resources in the physical space and the plan. This process reads the current load data from smart meters to calculate the effective operating rate of equipment, reads the entry and exit records of the real-name labor gates to calculate the working hour input rate, and reads the coordinate data from GPS locators to calculate the spatial location deviation. The formula for constructing the physical flow vector is as follows: ; In the formula, Represents the physical flow vector; This indicates the average load rate of the device within the current time slot; This represents the ratio of actual working hours to the standard working hours. This represents the normalized Euclidean distance between the actual geographic coordinates of the resource and the specified work area in the BIM model.

[0036] During the information flow vector construction process, the heterogeneous data vectorization processing unit parses unstructured engineering document logs to generate information flow vectors. This process uses keyword extraction technology to statistically analyze rectification records in supervision notices and calculates the circulation time of engineering acceptance documents. The formula for constructing the information flow vector is as follows: ; In the formula, Represents the information flow vector; This indicates the number of quality and safety rectification notices associated with the current time slot; This indicates the average delay time in the project acceptance document signing process.

[0037] The heterogeneous data vectorization processing unit is also equipped with a data normalization subroutine, which performs range transformation on the original values ​​of each dimension before generating the above vectors to eliminate dimensional differences and ensure that all vector elements are mapped to the closed interval [0, 1]. This normalization process ensures the mathematical fairness of the data contribution of each dimension in the subsequent entropy calculation process.

[0038] Through the above processing, the multi-source heterogeneous data mapping module transforms the originally discrete, heterogeneous, and difficult-to-quantify business data into a standardized state vector group that is mapped one-to-one with its corresponding time slot unit, providing a deterministic input data foundation for subsequent entropy value calculation.

[0039] Please see the appendix Figure 3 , Figure 3This is a data flow and logical architecture diagram of a dynamic performance status entropy calculation module according to an embodiment of the present invention. The dynamic performance status entropy calculation module is configured to receive standardized feature vectors from an upstream multi-source heterogeneous data mapping module and quantify the performance risk of resource slots using a hybrid entropy model. The module integrates a historical probability distribution update unit, a multi-dimensional weighted norm operation unit, and a deviation sensitivity response unit, which work together to output a single performance status entropy index.

[0040] The historical probability distribution update unit handles the traditional Shannon entropy component based on discrete events. This unit maintains a dynamically updated set of historical performance events, defining the discrete state space of the resource in past projects, including on-time delivery, minor delays, severe delays, quality defects, and default interruptions. For a specific resource, this unit employs a Bayesian update algorithm based on a time decay factor to calculate the posterior probability distribution of each discrete state event. This means that recent performance events contribute more significantly to the current probability distribution than longer-term events, ensuring that the probabilistic model accurately reflects the recent performance trend of the resource.

[0041] The multidimensional weighted norm computation unit is configured to process continuous cash flow feature vectors. Considering the varying degrees of impact of different indicators in the cash flow chain on performance risk, this unit introduces a diagonal weight matrix to linearly transform the cash flow vector, subsequently calculating the Euclidean norm of the transformed vector. This process compresses multidimensional financial data into scalar values ​​representing financial pressure, enabling the identification of hidden risks that appear liquid but exhibit abnormalities in key payment terms.

[0042] The Deviation Sensitivity Response Unit is used to handle execution deviations reflected in the physical flow and information flow data. This unit reads progress and efficiency data from the physical flow vector, compares it with the baseline plan set in the BIM model, and calculates the comprehensive deviation scalar at the current moment. This unit maps the deviation value to a non-linear exponential function, simulating the abrupt change characteristics of risks on the construction site. Specifically, when the deviation is within the allowable range, the entropy value increases slowly, but once the deviation exceeds a critical point, the entropy value will jump exponentially.

[0043] The dynamic performance state entropy calculation module performs a linear weighted aggregation of the outputs from the three processing units mentioned above to generate the final performance state entropy. The specific calculation model is as follows: ; In the formula, Representing resources In the time slot The performance state entropy; This represents the total number of categories of historical performance status events defined in the diagram; Indicates the first Similar to historical performance status events; Indicates the first The probability of such events occurring; Represents the cash flow vector; This represents a pre-defined diagonal weight matrix for the cash flow dimension, used to adjust the weights of different financial indicators on their contribution to risk. Represents the Euclidean norm operation for vectors; This represents the overall deviation scalar, whose value is derived from the physical flow vector. Positional deviation and information flow vector The number of rectification orders in the data is weighted and aggregated. The normalized adjustment coefficient representing the capital entropy component; The normalized adjustment coefficient representing the entropy component of the physical deviation; This represents the deviation sensitivity coefficient, used to control the response rate as the entropy value increases with deviation.

[0044] The dynamic performance status entropy calculation module is also equipped with parameter adaptive calibration logic. In the early stages of project operation, due to a lack of real-time data, the module automatically increases the weight of the first item (historical Shannon entropy); as the project progresses and the amount of real-time monitoring data increases, the module gradually increases... and The value of entropy allows the calculation of entropy to smoothly transition from historical experience to real-time on-site conditions.

[0045] Through the above calculation process, the dynamic performance state entropy calculation module transforms complex, multi-dimensional, and mixed qualitative and quantitative business data into a single numerical value that characterizes the degree of disorder in the system. This provides a unique metric for subsequent time-slot modulation and resource latching.

[0046] Please see the appendix Figure 4 , Figure 4 This is a schematic diagram of the control logic and data interaction of a time-slot adaptive modulation module according to an embodiment of the present invention. The time-slot adaptive modulation module is configured to monitor the dynamic change rate of the performance status entropy in real time, and accordingly perform physical-level elastic scaling control of the time-domain granularity of resources. The core processing logic of the time-slot adaptive modulation module includes an entropy gradient calculation subroutine, time-slot splitting execution logic, and time-slot merging callback logic.

[0047] The entropy gradient calculation subroutine is used to quantify the rate of change of resource performance risk over time. This subroutine reads the performance status entropy values ​​of the current time slot and the previous time slot, and derives the entropy gradient through discrete difference operations. The calculation of the entropy gradient aims to identify abrupt changes in risk trends, rather than focusing solely on the absolute magnitude of the risk. The formula for calculating the entropy gradient is defined as follows: ; In the formula, Representing resources In the time slot The entropy gradient value; This represents the performance state entropy of the current time slot; This represents the performance state entropy of the immediately preceding time slot; This indicates the time span of the current time slot.

[0048] The time slot splitting execution logic is configured to respond to high-frequency oscillations or sharp jumps in the entropy gradient. When the absolute value of the calculated entropy gradient exceeds a preset splitting threshold, the system determines that the resource has entered a period of sudden risk. At this time, the time slot adaptive modulation module activates the frequency domain refinement mechanism to perform physical splitting operations on future scheduled time slots that have not yet been executed. Specifically, this mechanism forcibly decomposes a single time slot object with an original time span of the initial standard time granularity into... A series of continuous micro-timeslot cells, in which The preset splitting factor is used. This splitting operation transforms the original linear storage units into pointer structures pointing to micro-slot arrays at the data structure level, thereby constructing a non-uniform time grid.

[0049] Simultaneously, the time-slot adaptive modulation module sends a synchronous acquisition command to the multi-source heterogeneous data mapping module and its underlying ETL interface via the reverse control link. This command forces an increase in the data acquisition frequency for that specific resource. The precision of the split micro-timeslots is adjusted to match the accuracy of the split. For example, when the time granularity is split from days to hours, the module drives the data reporting cycle of IoT devices to be adjusted synchronously to once per hour, ensuring that each micro-timeslot unit has independent real-time business data support, thereby eliminating monitoring blind spots during high-risk periods.

[0050] The time-slot merging callback logic is used to release computational resources after the risk regression has stabilized. This logic maintains a sliding time window, continuously monitoring the long-term trend of the entropy gradient. When consecutive... When the absolute value of the average entropy gradient over a period of time is lower than the preset merging threshold, the system determines that the fulfillment status of the resource has returned to stability.

[0051] At this point, the time-slot adaptive modulation module performs a reverse operation, re-aggregating adjacent micro-time-slot units into time-slot objects with standard time granularity, and simultaneously sending instructions to reduce the acquisition frequency of the underlying data. The merging threshold is set strictly lower than the splitting threshold to form a hysteresis comparison interval, preventing the system from frequently triggering splitting and merging oscillations near the critical point.

[0052] Through the above mechanism, the time-slot adaptive modulation module achieves dynamic modulation of the spatiotemporal resolution of resources. This mechanism changes the traditional model of resource management using a fixed time axis, enabling the system to automatically focus computing power and data acquisition bandwidth on high-risk resource nodes, thus achieving an adaptive balance between system load and monitoring accuracy.

[0053] The resource state latch control module is configured at the front end of the topology path matching and risk blocking module, serving as the physical layer access controller of the resource scheduling system. This module executes a strict multi-dimensional trust consensus protocol to verify the availability of each resource time slot and generates a corresponding binary latch status bit. This latch status bit, as a control signal at the system's lower level, directly determines whether a specific resource is visible and available to subsequent scheduling algorithms within a specific time window, thereby constructing a resource access barrier based on data trust.

[0054] The resource status latching control module first performs confidence level normalization. The module reads the cash flow vector, physical flow vector, and information flow vector generated by upstream mapping, and compresses and maps the multidimensional features of each vector to a closed interval of zero to one using a preset mapping function, thereby generating three independent real-time confidence scores. The cash flow confidence score represents the resource's current cash flow support capacity and credit level; the physical flow confidence score represents the actual status of on-site personnel and equipment input and operational efficiency; and the information flow confidence score represents the resource's compliance and cooperation in document circulation and rectification.

[0055] Based on the generated confidence scores, the resource state latch control module performs multi-dimensional confidence consensus logic operations. This operation presets minimum survival thresholds for three dimensions: funds, physical resources, and information. These minimum survival thresholds define the bottom-line indicators for allowing the system to operate normally. The resource state latch control module uses AND gate logic, requiring that scores for all dimensions simultaneously meet the threshold requirements to pass verification. Any weakness in any single dimension will trigger a rejection decision, thus preventing high-scoring items from masking low-scoring risk items in the weighted average algorithm.

[0056] The resource status latch control module determines the latch status bits of the resource time slot based on the above logic. The specific logic formula is as follows: ; In the formula, Representing resources In the time slot The latch status bit takes a Boolean value of 0 or 1; This indicates an indicative function that returns 1 if the condition within the parentheses is true, and 0 otherwise. This represents the logical AND operator. This indicates the confidence score of the funds; Indicates the physical confidence score; Indicates the confidence score of the information; This represents the minimum survival threshold in terms of funding. Represents the minimum survival threshold in the physical dimension; This represents the minimum survival threshold for an information dimension.

[0057] The physical significance of the latch status bit lies in achieving mandatory resource isolation. When the calculated latch status bit is 0, the system applies a mask to the resource slot at the database access layer or memory index layer, marking it as frozen. Although the resource slot in the frozen state physically exists, it is logically invisible to subsequent topology matching algorithms and cannot be retrieved or assigned to any task node. Only when the latch status bit is 1 is the resource slot marked as active, allowing it to enter the candidate resource pool and participate in optimal path planning.

[0058] In addition, the resource status latch control module supports receiving feedback signals from downstream risk blocking modules. When cascading risks are detected, it can forcibly overwrite the latch status bit to 0, thereby achieving preemptive locking of resources that will be put into use in future high-risk periods.

[0059] The topology path matching and risk blocking module is configured to perform resource scheduling based on minimizing global entropy increase at the system macro level, and to achieve predictive risk blocking by calculating the network propagation of risk flux during the execution phase. This module integrates a graph theory optimization solver, a risk flux calculation engine, and a cascading effect blocking controller.

[0060] The topology path matching and risk prevention module first constructs a directed acyclic graph (DAG) of the project tasks. This graph structure decomposes the engineering project into a series of discrete task nodes, using directed edges to represent strict pre- and post-process dependencies between tasks. The module maps candidate resource slots that have been filtered by the aforementioned resource state latching control module and whose latching state bits are active (i.e., value 1) to feasible solution sets for each task node in the graph.

[0061] Based on this, the module execution path optimization algorithm aims to find a resource allocation combination that minimizes the cumulative performance entropy of the entire project network while satisfying the constraints of the total project duration and total cost budget. The mathematical expression of this optimization objective function is as follows: ; In the formula, This represents the target value for the increase in total entropy of the system to be optimized. This represents the set of candidate resource allocation schemes, where each element is a candidate resource allocation scheme. Indicates resources In the time slot Assigned to task nodes ; This represents the fulfillment state entropy of the corresponding resource time slot. This objective function ensures that the system not only focuses on the local optima of a single node, but also pursues the maximization of the macroscopic orderliness of the entire supply chain network.

[0062] After the project enters the dynamic execution phase, the topology path matching and risk prevention module activates the risk flux calculation engine. This engine compares the actual running status of task nodes with their planned status in real time. When it detects that the entropy value of a node in actual execution exceeds its predicted entropy value in the matching phase, it determines that the node has generated entropy overflow, i.e., risk flux. Risk flux represents the amount of energy that local disorder spreads to other parts of the system. The formula for calculating risk flux is as follows: ; In the formula, Represents task node The resulting risk flux; This represents the actual performance status entropy calculated based on real-time monitoring data; This represents the preset baseline prediction entropy value during the initial matching phase.

[0063] Based on the calculated risk flux, the module utilizes a cascading effect to block the controller's deduction of the risk propagation path in the DAG network. This controller calculates the impact of the risk flux of upstream nodes on the future states of downstream related nodes based on the strength of dependencies between tasks and time intervals. This calculation simulates the decay and accumulation process of risk in the supply chain network, used to update the predicted entropy values ​​of downstream nodes.

[0064] The new predicted entropy value after the downstream nodes are affected is calculated as follows: ; In the formula, Indicates downstream node The updated predicted entropy value after being affected by upstream risk transmission; Represents a node The initial predicted entropy value; Represents a node The set of all preceding dependent nodes; The target downstream task node for the current calculation; For any preceding task node in the set of all preceding dependent nodes; Indicates from node To the node The correlation coefficient is determined by the degree of process dependence. For the preceding task node The resulting risk flux; Represents a node The planned start time; Represents a node The planned end time; This represents the time-damping factor, used to simulate the natural decay effect of risk over time. is the base of the natural logarithm.

[0065] The cascade effect blocking controller performs dynamic blocking decisions based on the updated predicted entropy value. When the calculated new predicted entropy value of the downstream node causes its corresponding predicted entropy gradient value to exceed the splitting threshold set by the aforementioned time slot adaptive modulation module, the system determines that the downstream node faces an imminent risk of systemic collapse.

[0066] At this point, the module immediately generates a high-priority blocking command and sends it back to the resource state latching control module. This command forcibly blocks the downstream node. The latch status bits of the originally intended matching resource are overwritten to the locked state (i.e., set to 0), thereby physically cutting off the intervention path of the high-risk resource. Simultaneously, the module triggers a rematch interrupt signal, initiating a process targeting the node. The alternative resource optimization process aims to complete resource replacement and path reconstruction before the actual occurrence of risks.

[0067] 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 construction industry resource matching and performance early warning system based on business data fusion, characterized in that: The system includes: The multi-source heterogeneous data mapping module is configured to build a digital basic model of construction supply chain resources. It discretizes labor, equipment and material resources in the time dimension to construct a resource-time slot matrix, and converts the capital flow data, physical flow data and information flow data collected from external interfaces into standardized feature vectors respectively. The dynamic performance status entropy calculation module is communicatively connected to the multi-source heterogeneous data mapping module and is configured to receive the feature vector and calculate the performance status entropy of each resource time slot. The performance status entropy is used to quantify the system disorder and performance risk value of a specific resource within a specific time window. The time slot adaptive modulation module is connected to the dynamic performance state entropy calculation module and the external data acquisition interface respectively. It is configured to monitor the entropy gradient of the performance state entropy in real time, and dynamically control the physical granularity of the resource time slot to perform splitting or merging operations according to the change amplitude of the entropy gradient, and synchronously send instructions to the external data acquisition interface to adjust the sampling frequency of the underlying data. The resource status latch control module is configured at the front end of the topology path matching and risk blocking module. It is configured to execute a confidence consensus protocol based on multidimensional data to generate binary latch status bits. The latch status bits determine the visibility and availability of resource slots in the system. The topology path matching and risk blocking module is configured to construct a directed acyclic graph of project tasks. In the resource pool that has been filtered by the resource status latch control module and whose latch status bit is active, it finds the resource configuration path with the minimum increase in total system entropy and is configured to calculate the network propagation of risk flux. When it is predicted that a downstream node is facing high risk, it directly modifies the latch status bit of the associated resource to cut off the risk propagation path.

2. The construction industry resource matching and performance early warning system based on business data fusion according to claim 1, characterized in that, The multi-source heterogeneous data mapping module includes a heterogeneous data vectorization processing unit, which is configured as follows: Payment and credit records associated with specific resources are extracted from the financial system to generate a fund flow vector. The features of the fund flow vector aggregation include the normalized average bill of exchange term, the normalized historical cumulative payment default frequency, and the inverse of the account liquidity ratio corresponding to the current time slot. Physical flow vectors are generated based on real-time data collected by on-site sensor networks. The features of the aggregated physical flow vectors include the average load rate of the equipment in the current time slot, the ratio of actual working hours to standard working hours, and the normalized Euclidean distance between the actual geographic coordinates of the resources and the work area specified by the model. The engineering document log is parsed to generate an information flow vector. The features of the aggregated information flow vector include the number of quality and safety rectification notices associated with the current time slot and the average lag time of the engineering acceptance document signing process.

3. The construction industry resource matching and performance early warning system based on business data fusion according to claim 2, characterized in that, The dynamic performance state entropy calculation module calculates the performance state entropy using a hybrid entropy model, and the calculation process includes: Using a Bayesian update algorithm based on time decay factor, the posterior probability distribution of historical performance status events is calculated, and the Shannon entropy component is calculated accordingly. A diagonal weight matrix for the capital flow dimension is introduced to perform a linear transformation on the capital flow vector, and the Euclidean norm of the transformed vector is calculated as the capital entropy component. A comprehensive deviation scalar is calculated and mapped using a nonlinear exponential function as a physical deviation entropy component. The comprehensive deviation scalar is a weighted composite of the position deviation in the physical flow vector and the number of rectification orders in the information flow vector. The Shannon entropy component, the capital entropy component, and the physical deviation entropy component are linearly weighted and aggregated to obtain the performance status entropy.

4. The construction industry resource matching and performance early warning system based on business data fusion according to claim 3, characterized in that, The time slot adaptive modulation module includes an entropy gradient calculation subroutine and time slot splitting execution logic; The entropy gradient calculation subroutine is configured to read the performance status entropy of the current time slot and the previous time slot, and calculate the entropy gradient through discrete difference operation; The time slot splitting execution logic is configured such that when the absolute value of the entropy gradient exceeds a preset splitting threshold, the single time slot object with the original time span as the initial standard time granularity is decomposed into multiple consecutive micro-time slot units, and a synchronous acquisition command is sent to the multi-source heterogeneous data mapping module through the reverse control link to forcibly increase the data acquisition frequency for the specific resource in order to match the accuracy of the split micro-time slots.

5. The construction industry resource matching and performance early warning system based on business data fusion according to claim 4, characterized in that, The time slot adaptive modulation module also includes time slot merging callback logic; The time slot merging callback logic is configured to maintain a sliding time window. When the absolute value of the average entropy gradient of multiple consecutive periods within the sliding time window is lower than the preset merging threshold, the adjacent micro time slot units are re-aggregated into time slot objects with standard time granularity, and instructions are sent synchronously to reduce the acquisition frequency of the underlying data. The set value of the merging threshold is less than the set value of the splitting threshold to form a hysteresis comparison interval.

6. The construction industry resource matching and performance early warning system based on business data fusion according to claim 1, characterized in that, The resource state latch control module is configured to execute the following confidence consensus protocol to determine the latch state bits: The capital flow vector, the physical flow vector, and the information flow vector are mapped to capital confidence scores, physical confidence scores, and information confidence scores, respectively. Preset minimum survival thresholds for the financial, physical, and informational dimensions respectively; Using a logical AND operation, the latched state bit is marked as active only when the financial confidence score, the physical confidence score, and the information confidence score are all greater than or equal to their respective minimum survival thresholds; otherwise, the latched state bit is marked as frozen, and a masking code is applied to the resource time slots in the frozen state at the system logic layer.

7. The construction industry resource matching and performance early warning system based on business data fusion according to claim 1, characterized in that, The topology path matching and risk blocking module is configured to run a path optimization algorithm. The objective function of the path optimization algorithm is configured to find a resource allocation combination that minimizes the sum of the performance state entropy of all selected resource slots in the candidate resource allocation scheme set.

8. The construction industry resource matching and performance early warning system based on business data fusion according to claim 1, characterized in that, The topology path matching and risk blocking module includes a risk throughput calculation engine; The risk flux calculation engine is configured to compare the actual running status and planned status of task nodes in real time. When it is detected that the performance status entropy of a certain node in actual execution exceeds the baseline predicted entropy value of the node in the initial matching stage, the difference between the two is calculated as the risk flux. The risk flux represents the amount of energy that local disorder spreads to the system.

9. The construction industry resource matching and performance early warning system based on business data fusion according to claim 8, characterized in that, The topology path matching and risk blocking module also includes a cascade effect blocking controller, which is configured as follows: Calculate the impact of the risk flux of the upstream node on the future state of the downstream associated node to update the predicted entropy value of the downstream node; The updated prediction entropy value of the downstream node is obtained by superimposing the initial prediction entropy value of the downstream node with the propagation influence value of all preceding dependent nodes; The calculation of the transmission impact value is based on the risk flux of the upstream dependent node, the correlation impact coefficient from the upstream node to the downstream node, and the natural exponential decay term based on the time damping factor. The time variable of the natural exponential decay term is the difference between the planned start time of the downstream node and the planned end time of the upstream node.

10. The construction industry resource matching and performance early warning system based on business data fusion according to claim 9, characterized in that, The cascade effect blocking controller is also configured to: When the predicted entropy gradient value calculated based on the updated predicted entropy value exceeds the splitting threshold set by the time slot adaptive modulation module, a blocking command is generated and fed back to the resource status latching control module. In response to the blocking command, the resource status latch control module forcibly overwrites the latch status bit of the originally matched resource of the downstream node to the locked state, and triggers a rematch interrupt signal to start the alternative resource optimization procedure.

Images