Assembly type electromechanical facility linkage method based on digital configuration and BIM mapping

Abstract

The invention belongs to the technical field of computer data processing and automatic control, and particularly discloses and provides a digital configuration and BIM (Building Information Modeling) mapping fabricated electromechanical facility linkage method, which comprises the following steps of: analyzing a building information model and digital configuration, and constructing an initial weighted mapping network; initial virtual pressure of each node is calculated, conduction and superposition of the pressure in the network are iteratively simulated, and a dynamic pressure distribution diagram representing global risks is generated; and when an anomaly is detected, reverse tracing is carried out along a path with the maximum pressure contribution degree gradient so as to locate a risk source. Furthermore, various candidate control strategies are deduced through simulation, scoring is carried out from multiple dimensions, and the optimal strategy is selected. In addition, by comparing simulation and actual execution effects, deviation data is utilized to adaptively optimize the weighted mapping network and simulation model parameters. According to the method, accurate prediction, rapid traceability, optimal decision and continuous self-learning of the risk of the complex facility system are realized, and the intelligence and reliability of operation and maintenance are improved.

Description

An Interlocking Method for Prefabricated Mechanical and Electrical Facilities with Digital Configuration and BIM Mapping Technical Field

[0001] The present invention belongs to the technical field of computer data processing and automatic control, and relates to an interlocking method for prefabricated mechanical and electrical facilities with digital configuration and BIM mapping. Background Art

[0002] A large number of complex mechanical and electrical facilities are integrated inside prefabricated buildings, such as air conditioning systems, water supply and drainage systems, power supply and distribution systems, etc. The stable operation of these facilities is crucial for the overall function and safety of the building. In order to achieve efficient operation and maintenance and intelligent control of these facilities, building information modeling (BIM) technology is usually used to manage the spatial location and physical connection relationship of the facilities, and a digital configuration system or a supervisory control and data acquisition (SCADA) system is used to monitor their operating parameters in real time. Integrating the static design data carried by BIM with the dynamic operation data generated by the configuration system and realizing intelligent interlocking control between facilities based on this is a core research direction in the fields of smart buildings and digital twins.

[0003] In the prior art, to achieve the interlocking control between BIM and the configuration system, a relatively direct point-to-point mapping method is usually adopted. Specifically, technicians manually create a mapping table to statically bind the ID of a certain device component in the BIM model with one or more monitoring data point tags in the configuration system. When interlocking needs to be achieved, fixed hard-coded rules are pre-written in the system, such as "when the real-time value of the temperature sensor in a certain area is higher than the set upper limit, trigger an instruction to start the corresponding air conditioning fan in that area". When the system detects a data change that meets the rule conditions, it finds the corresponding execution device by referring to the static mapping table and issues a preset control instruction.

[0004] However, the above prior art solutions have obvious technical defects in practical applications. First, the data fusion method based on the static mapping table is very fragile. Once on-site equipment is replaced, the circuit is modified, or the sensor position is adjusted, the manually maintained mapping table often cannot be updated in a timely manner, resulting in incorrect data association, and then causing the failure or misjudgment of the control logic. Second, the hard-coded rules that rely on the threshold of a single data point to trigger have a very narrow decision-making perspective and cannot perceive and evaluate the chain reaction that a control action may have on the entire interconnected mechanical and electrical system network. It is easy to have a situation where local control measures do not consider the interlocking impact on the overall system, which belongs to local optimization rather than global optimization, and may even cause more extensive or hidden systemic problems in order to solve a local problem. Summary of the Invention

[0005] In order to overcome the above-mentioned defects of the prior art and to achieve the above objectives, the present invention proposes the following technical solution: a method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping, comprising: S1, acquiring the physical connection relationship of equipment in the building information model and the real-time monitoring data in the digital configuration system, and establishing an initial correspondence relationship based on the unique identifier or tag of the equipment, and generating an initial weighted mapping network.

[0006] S2. Based on the initial weighted mapping network, calculate the deviation of the real-time monitoring data of each monitoring point from its target value range, generate the initial virtual pressure value of each node, and use the initial weighted mapping network to transmit and superimpose the pressure to generate a dynamic pressure distribution map representing the global risk situation.

[0007] S3. Identify abnormal pressure nodes in the dynamic pressure distribution map and perform reverse pressure gradient tracing to determine the pressure root cause device. Based on this, extract a local network subgraph containing the pressure root cause device and its affected neighbor devices.

[0008] S4. Call the physical model equations that match the device types in the local network subgraph, and combine them with the topological connections of the local network subgraph to construct a local simulation sandbox. In the local simulation sandbox, deduce multiple candidate control strategies in parallel to generate future state deduction results.

[0009] S5. Quantitatively score the future state projection results according to the multi-dimensional security evaluation rules, select the candidate control strategy with the highest score as the optimal control command, and collect the actual response data after the command is executed.

[0010] S6. Compare the actual response data with the future state projection results to generate policy effect deviation data. Based on this, adjust the weight coefficients in the initial weighted mapping network to generate an updated weighted mapping network for subsequent control cycles.

[0011] Compared with the prior art, the beneficial effects of the present invention are as follows: (1) By constructing a dynamic mapping network and introducing a self-learning feedback mechanism, the present invention enables the system to automatically discover and quantify the inconsistencies between the BIM static model and the configuration dynamic data, and dynamically adjust the credibility of the data fusion relationship. This makes the foundation of data processing more solid and reliable, and the system can autonomously adapt to changes in field equipment or sensor aging, thereby improving the robustness of data processing and the stability of long-term operation of the entire linkage control system.

[0012] (2) This invention innovatively transforms equipment operating deviations into virtual pressures that can be transmitted and superimposed in the physical topology network, and generates a global dynamic pressure distribution map based on this. This data processing method enables the system to gain insight into the systemic and cascading risks caused by local anomalies from a macroscopic perspective, realizing a leap from processing isolated data points to analyzing the global data potential field, and enhancing the system's ability to predict potential faults and the accuracy of situational awareness.

[0013] (3) This invention proposes a decision-making deduction and safety verification method based on a local simulation sandbox. This method uses a dynamically constructed lightweight mathematical model to perform rapid and parallel virtual execution and evaluation of multiple control strategies, ensuring that the final issued command is the optimal solution obtained by weighing multiple constraints such as safety, energy efficiency, and equipment impact. This combination of data processing and simulation provides deterministic computational assurance for the safety of control commands while ensuring the efficiency of automated decision-making, thus solving the problems of high decision-making risk and reliance on manual intervention in traditional automated control.

[0014] (4) This invention designs a complete closed-loop data processing flow from data acquisition, processing, modeling, decision-making to feedback optimization. By continuously comparing the actual execution results with the simulation predictions, the system can continuously correct the weights of its internal mapping network and the parameters of the physical model. This self-iterative and optimized data processing mechanism enables the system's intelligence level and the accuracy of its simulation of the physical world to continuously increase over time, ultimately realizing an intelligent linkage control system that can evolve autonomously and is highly adaptable. Attached Figure Description

[0015] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 is a schematic diagram of the implementation steps of the method of the present invention. Detailed Implementation

[0017] The technical solutions of 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.

[0018] Please refer to Figure 1. The present invention proposes a method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping, which includes: S1, obtaining the physical connection relationship of equipment in the building information model and the real-time monitoring data in the digital configuration system, and establishing an initial correspondence relationship based on the unique identifier or tag of the equipment to generate an initial weighted mapping network.

[0019] In a preferred embodiment, generating an initial weighted mapping network includes: parsing the three-dimensional data structure of the building information model, extracting the physical connection relationships between electromechanical equipment nodes and equipment endpoints, and constructing an equipment physical topology map reflecting the spatial layout of the facility; traversing the monitoring point list in the digital configuration system, and establishing dynamic associations between real-time data streams and nodes in the equipment physical topology map based on the corresponding rules of equipment unique identifiers or tag numbers; extracting physical attribute parameters of physical connection relationships, quantifying the transmission influence coefficient of each connection edge according to the physical attribute parameters, and assigning this as a weight value to the corresponding edge of the equipment physical topology map to form an initial weighted mapping network.

[0020] Specifically, this step aims to construct a digital topology foundation capable of supporting dynamic risk transmission calculations. Traditional BIM models only contain static geometric and attribute information, while digital configuration systems such as SCADA only record discrete real-time data, lacking logical connection between the two. This step achieves deep integration of the two through the following sub-steps: Step 1: Constructing the Equipment Physical Topology Diagram: The system first loads the Building Information Model (BIM) data of the target building, such as IFC standard format files or Revit native files. The parsing engine traverses all MEP objects in the model, extracting the 3D geometric data of MEP equipment (nodes) and their associated pipes, cable trays, or lines (connection edges). In practice, the system analyzes whether there are physical connectivity paths between equipment by identifying the connection ports and pipe segments of the equipment. Based on graph theory algorithms, MEP equipment is instantiated as "topology nodes," and connecting pipes are instantiated as "undirected edges," thereby constructing an equipment physical topology diagram reflecting the actual spatial layout on site. During this process, the system also simultaneously extracts the physical attribute parameters of each connection edge, including but not limited to pipe diameter, physical length, medium type (such as water, gas, and current), and material roughness.

[0021] The second step is to establish dynamic data mapping: The system connects to the field digital configuration system via industrial communication protocols such as OPCUA, MQTT, or Modbus to obtain real-time monitoring data streams. To bind BIM nodes to real-time data, the system executes an identifier mapping procedure: Identifier Extraction: Extracts the unique identifier of each device node from the BIM model, such as a GUID or design tag number, while simultaneously traversing the list of measurement point labels in the digital configuration system.

[0022] Rule matching: Using preset regular expressions or lookup tables, the device identifier is matched with the measurement point label. For example, the BIM node “AHU_F1_01” is automatically mapped to the configuration labels “AHU_01_RunStatus” and “AHU_01_Alarm”.

[0023] Attribute binding: After a successful match, the real-time data stream is attached as a dynamic attribute to the corresponding topology node, enabling the static physical topology graph to have real-time status awareness.

[0024] Step 3: Quantifying Connection Weights: To simulate the transmission characteristics of risks such as stress, overload, or failure in the network, physical connection relationships need to be converted into digitized mathematical weights. Based on the physical attribute parameters extracted in Step 1, the transmission influence coefficient of each edge is calculated and assigned as a weight value to the corresponding edge in the device's physical topology diagram. The specific weight calculation logic can be as follows: Positive correlation factor: The cross-sectional size of the connection edge (such as pipe diameter or wire diameter) is a positive indicator of transmission capacity (the larger the size, the smaller the transmission resistance or the greater the influence), and is assigned a higher weight base; Negative correlation factor: The physical length of the connection edge is an attenuation indicator of transmission capacity (the longer the distance, the weaker the transmission influence), and is used as the attenuation coefficient of the weight.

[0025] For example, for a large-diameter main water supply pipe with a short distance, the system calculates a higher weight value (such as...). This indicates that the path can transmit pressure fluctuations almost without loss; while for a slender branch, a lower weight value is calculated (e.g. ).

[0026] It should be noted that the transmission influence coefficient The following exemplary formula can be used for quantification and normalization:

[0027] In the formula, The normalized propagation influence coefficient represents the connection edge from device node i to device node j; It is the cross-sectional area of ​​the connecting edge from device node i to device node j, which is automatically extracted or calculated from the geometric properties of the BIM model and used as a positive correlation factor. It is the physical length of the connection edge from device node i to device node j, which is directly read from the geometric properties of the BIM model as a negative correlation factor; It is a preset distance attenuation index, usually set to 1 or adjusted based on empirical data, used to adjust the contribution of physical length to the attenuation. It is a conduction reference coefficient related to the type of connection medium m, such as the main water supply pipe that can be preset. Branch duct High-voltage cables This is to reflect the inherent differences in conductivity between different media; This represents a preset dimensional correction factor, which cancels out the physical dimensions of the expression, ensuring that the input to the normalization function is a dimensionless pure number. The value of this factor is constant at 1, and its unit is [length]. (x-2) ; The normalization function, in this embodiment, is specifically implemented as linear function normalization (Min-Max Scaling), used to map the calculation results to the [0,1] interval. It should be emphasized that this formula is merely an exemplary implementation; any other functional relationship that reflects the influence of physical properties on conduction falls within the scope of this invention.

[0028] After the above processing, the system finally generates an initial weighted mapping network. This network not only includes the connection relationships of the devices, but also quantified transmission capability parameters and real-time operating status data, providing a physically meaningful data structure foundation for the anomaly transmission calculation in the subsequent step S2.

[0029] S2. Based on the initial weighted mapping network, calculate the deviation of the real-time monitoring data of each monitoring point from its target value range, generate the initial virtual pressure value of each node, and use the initial weighted mapping network to transmit and superimpose the pressure to generate a dynamic pressure distribution map representing the global risk situation.

[0030] In a preferred embodiment, generating the initial virtual pressure value for each node includes: acquiring the real-time monitoring data corresponding to each monitoring point and its target value range stored in the device attribute library; calculating the normalized value of the real-time monitoring data deviating from the target value range to obtain the deviation degree; acquiring the device criticality weight associated with the monitoring point and the duration of the current abnormal state; and calculating the initial virtual pressure value by positively weighting the deviation degree and the device criticality weight, and then performing logarithmic processing on the duration of the current abnormal state.

[0031] Specifically, the engineering objective of calculating the initial virtual pressure value for each monitoring point in this invention is to transform heterogeneous, multi-dimensional equipment operation data into a standardized, quantitative indicator that can characterize the operational risk level of a single piece of equipment. This indicator not only reflects the severity of parameter deviations but also incorporates the importance of the equipment itself and the persistent impact of abnormal states, providing a comparable input source for subsequent global risk situation transmission. The detailed engineering implementation steps are as follows: Step 1: Perform data acquisition and benchmark establishment steps: At a preset polling cycle, such as 2 to 10 seconds, actively acquire the real-time measurement value of each monitoring point through the industrial bus protocol or API interface. After acquiring the real-time measurement value, immediately retrieve the preset target value range corresponding to the monitoring point from the equipment attribute database associated with the building information model. This range is a numerical range containing upper and lower limits, defining the ideal operating boundary of the equipment parameters under normal operating conditions.

[0032] The second step is to perform dimensionless calculation of the deviation: The engineering goal is to eliminate differences caused by different physical dimensions and numerical ranges, so that the deviations of different parameters such as pressure, temperature, and flow rate can be compared on a unified scale. When the system determines that the real-time measured value exceeds the preset target value range, it initiates the calculation of the normalized deviation. The calculation formula is as follows:

[0033] In the formula, It represents the normalization deviation and is a dimensionless scalar. Represents real-time measured values; and These represent the upper and lower limits of the preset target value range, respectively. The boundary value represents the interval of the target value closest to the real-time measured value, i.e., when hour ,when hour If the real-time measured value is within the target value range, then the normalized deviation is... It is zero.

[0034] The third step is to introduce a risk weighting factor: The engineering objective is to reflect the differences in the impact of different devices and different durations of anomalies on the overall system security. The system reads the preset criticality weights of the devices associated with the current monitoring point from the device attribute database. This weight is a value pre-set by operations and maintenance experts based on the device's importance in the system, redundant configuration, and potential failure impact, typically between 1.0 and 5.0. Simultaneously, the system's internal anomaly monitor records the duration of the current anomaly where the current parameter deviates from the behavior. This timer starts when the anomaly first occurs and resets to zero after the parameters return to normal and stabilize for a period of time.

[0035] Step 4: Perform multi-factor fusion calculations to generate the final initial virtual stress value: The engineering objective of this step is to combine all the above risk factors into a comprehensive risk indicator. The system uses a preset stress calculation function to perform nonlinear combination of each factor, as shown in the formula:

[0036] In the formula, This is the initial virtual pressure value to be output; To obtain the preset criticality weights, a gradation is usually set, such as 1 to 5. An example setting is: core chiller. Main circulation pump Terminal fan coil units Ordinary temperature sensor ; The duration of the current anomaly; This is a time scale constant used to standardize the duration of anomalies. For example, for a power system requiring a response time in seconds, it can be set to... s; The preset time sensitivity coefficient is used to adjust the degree of influence of the duration of the anomaly on the result. After testing, it was found that... or It can keep the S-value between 1.0 and 50.0 in most typical abnormal scenarios, which is a relatively ideal range; It is a natural logarithmic function. This formula ensures that the initial virtual pressure value amplifies non-linearly with increasing deviation, equipment importance, and anomaly duration, thus more accurately simulating the cumulative effect of risk in actual engineering. This initial virtual pressure value will serve as input data for the next stress transmission calculation.

[0037] In a further preferred embodiment, an initial weighted mapping network is used to transmit and superimpose pressure to generate a dynamic pressure distribution map representing the global risk situation. This includes: obtaining the transmission influence coefficient assigned to each connection edge in the initial weighted mapping network; multiplying the comprehensive virtual pressure value of the upstream node by the transmission influence coefficient of the connection edge to obtain the attenuated pressure value transmitted to the downstream node; at each device node, summing the node's own initial virtual pressure value with all attenuated pressure values ​​transmitted from the upstream nodes to generate a new round of comprehensive virtual pressure value for the node, and the comprehensive virtual pressure values ​​of all nodes constitute the dynamic pressure distribution map.

[0038] Specifically, the engineering purpose of transmitting and superimposing the initial virtual pressure value in this invention is to simulate the process of the propagation and accumulation of faults or anomalies along physical media such as pipelines or circuits in the physical world. It transforms isolated equipment risk points into a global situational view that reflects systemic and cascading risks, i.e., a dynamic pressure distribution map, thereby enabling the system to anticipate and locate global hidden dangers caused by local problems. The detailed engineering implementation steps are as follows: Step 1: Initialize pressure state: The system first calculates the initial virtual pressure value for each device node j in the network. This value is calculated based on the node's real-time data deviation, equipment criticality weight, and anomaly duration, serving as an inherent, originating risk measure for that node. Simultaneously, the system will incorporate the comprehensive virtual stress value of each node. Initialize to this initial value, that is , where the superscript (0) represents the 0th iteration.

[0039] The second step is to perform iterative calculations of the global pressure: The engineering goal of this step is to achieve a stable and convergent pressure distribution across the entire network through multiple iterations, accurately reflecting the combined impact of all risk sources. The system iterates through all device nodes j in the network to perform calculations. In each iteration (taking the (t+1)th iteration as an example), a new round of comprehensive virtual pressure value is calculated for each device node. Each is determined by its own initial virtual pressure value. It is the sum of all pressure components received from upstream neighbors. The calculation formula is:

[0040] In the formula, This represents the new round of comprehensive virtual pressure value calculated for downstream device node j in this iteration; The forgetting factor or damping coefficient, representing the range (0,1), indicates how much of the previous stress is retained. For example, setting... This means that in each iteration, a node retains 85% of its previous pressure value and incorporates 15% of the new pressure impact on top of that. This is a stable setting. This represents the total virtual pressure value of upstream node i at the t-th iteration; It is the initial virtual pressure value of device node j itself, which remains unchanged throughout the entire iteration process; Represents the set of all upstream neighbors belonging to node j. Sum the values ​​at node i; It is the combined virtual pressure value calculated by the upstream neighbor node i in the previous iteration; It is the transmission influence coefficient connecting upstream node i and downstream node j, which is obtained directly from the initial weighted mapping network.

[0041] Step 3: Determine convergence and output the results: This iterative process will be repeated until the change in the overall virtual pressure value of all nodes in the entire network between two adjacent iterations (e.g., The sum or maximum value of all virtual pressure values ​​(i.e., the sum or maximum value of all virtual pressure values) is less than a preset convergence threshold (e.g., one ten-thousandth), or the preset maximum number of iterations is reached. After convergence, the final, stable comprehensive virtual pressure values ​​of all device nodes are aggregated to form a data set with the device's unique identifier as the key and its comprehensive virtual pressure value as the value. This data set is the dynamic pressure distribution map of the entire network, which will serve as the direct basis for subsequent risk localization and decision analysis.

[0042] It should be noted that the "virtual pressure value" proposed in this invention is not an abstract concept detached from physical reality, but rather a comprehensive risk quantification indicator used to characterize the deviation of the operating state of electromechanical equipment from its design safe range. This indicator is calculated by integrating multiple technical parameters such as real-time physical measurements (e.g., temperature, pressure, flow rate), the importance of the equipment in the physical system, and the physical duration of abnormal states. Its value directly reflects the potential operational risks borne by the equipment. Similarly, the "dynamic pressure distribution map" is a data structure that carries the "virtual pressure values" of all nodes in the entire system. It reconstructs the objective process of risk transmission and accumulation along physical connections such as pipelines and lines in the physical world in the digital space. This invention utilizes these technical means to transform complex, multi-dimensional equipment operating data into a calculable and traceable unified model, thereby achieving fault location and intelligent control of physical facilities. Its essence is to solve a specific technical problem.

[0043] S3. Identify abnormal pressure nodes in the dynamic pressure distribution map and perform reverse pressure gradient tracing to determine the pressure root cause device. Based on this, extract a local network subgraph containing the pressure root cause device and its affected neighbor devices.

[0044] In a preferred embodiment, identifying abnormal pressure nodes in a dynamic pressure distribution map and performing reverse pressure gradient tracing to determine the pressure root cause device includes: identifying abnormal pressure nodes in the dynamic pressure distribution map whose comprehensive virtual pressure value exceeds a preset alarm threshold; calculating the virtual pressure difference between the abnormal pressure node and all directly connected upstream nodes, generating a pressure change gradient set; selecting the connection path with the largest gradient value from the pressure change gradient set as the main conduction path; setting the upstream node of the main conduction path as the new current node, and repeating the steps of calculating the pressure change gradient and selecting the main conduction path until the virtual pressure value of the current node is lower than the preset root cause determination threshold, and finally determining the node that meets the conditions as the pressure root cause device.

[0045] Specifically, the engineering objective of reverse pressure gradient tracing along the physical topology network of the equipment in this invention is to automatically locate the root cause of the system's cascading anomalies. This method simulates the reverse propagation of energy or influence along the path of greatest resistance or strongest force in the physical world, peeling back the layers from the surface problem points to the core of the problem—the pressure-generating equipment. The detailed engineering implementation steps are as follows: Step 1: Initialize the tracing starting point: When the comprehensive virtual pressure value of any node in the dynamic pressure distribution map first exceeds a preset alarm threshold, the node is locked by the system and marked as an abnormal pressure node, serving as the starting point for this reverse tracing algorithm. The system sets this abnormal pressure node as the initial current node i and reads its complete connection information from the physical topology network of the equipment.

[0046] The second step involves repeatedly performing the following operations, using the current node i as the core: Neighborhood search and gradient quantization: The system queries the physical topology network of the devices to identify the set of all upstream device nodes j directly connected to the current node i. For each upstream node j, the system calculates the gradient of its pressure contribution to the current node i. The engineering significance of this gradient lies in quantifying the extent to which the pressure of upstream node j "contributes" to the total pressure of current node i. Its calculation formula is:

[0047] In the formula, Represents upstream equipment To the current device The pressure contribution gradient; It is the comprehensive virtual pressure value of upstream node j; It is the transmission influence coefficient connecting upstream node j and current node i, which is obtained directly from the initial weighted mapping network.

[0048] Path decision: The system compares all calculated pressure contribution gradients and selects the upstream node with the largest gradient value. and connect the path This has been identified as the main transmission path for this round of tracing.

[0049] Node update and termination judgment: The system will update the upstream node of the main propagation path. Update to the new "current node i" and complete one iteration. Then, perform termination condition checks on the new current node: Condition 1 (Pressure Threshold): Check the comprehensive virtual pressure value of the new current node. Whether it has fallen below the preset root cause determination threshold, which usually represents the normal pressure fluctuation range of the system.

[0050] Condition 2 (Topology Boundary): Check if the new current node has no upstream connections.

[0051] Step 3: Identify and Output the Root Cause Device: The iterative tracing process ends when any of the above termination conditions is triggered. The system will formally identify the node that caused the loop termination—that is, the last node whose comprehensive virtual pressure value is still higher than the root cause determination threshold and has an upstream connection—as the pressure root cause device for this abnormal event. The information of this pressure root cause device will be passed to the subsequent decision-making and simulation modules as the core basis for formulating control strategies.

[0052] S4. Call the physical model equations that match the device types in the local network subgraph, and combine them with the topological connections of the local network subgraph to construct a local simulation sandbox. In the local simulation sandbox, deduce multiple candidate control strategies in parallel to generate future state deduction results.

[0053] In a preferred embodiment, constructing a local simulation sandbox includes: matching each device in the local network subgraph with a corresponding core physical law equation from a model library storing various device physical models; obtaining the real-time operating parameters of each device in the local network subgraph as initial conditions for the core physical law equations; and converting the topological connection relationship of the local network subgraph into boundary constraints and flow conservation constraints between the core physical law equations to complete the initialization of the local simulation sandbox.

[0054] Specifically, the engineering objective of constructing a local simulation sandbox in this invention is to dynamically and automatically generate a lightweight, computationally efficient mathematical model that reproduces the core physical dynamics within the current anomaly-affected area. This aims to avoid the enormous computational overhead and time delay of performing detailed simulations of the entire system, thereby providing a high-fidelity virtual testing environment for the rapid, real-time, and safe simulation of subsequent control strategies. The detailed engineering implementation steps are as follows: Step 1: Execute intelligent matching and equation selection from the model library: After receiving the local network subgraph determined in the previous step, the system begins to analyze the type and attributes of each device node contained within it. It traverses all devices in the subgraph, such as centrifugal pumps, regulating valves, heat exchange coils, etc., and uses these device types as query indexes to search in a pre-set simplified physical model library. This model library pre-stores the core physical law equations for various types of electromechanical equipment. For example, for fluid pipelines, the library stores simplified forms of Bernoulli's equations or Darcy's formulas; for water pumps, it stores polynomial fitting equations for their performance curves. The system selects a set of core physical law equations that are computationally concise for each device and connecting pipe segment in the local network subgraph through matching.

[0055] The second step is the parameterization and state initialization of the execution model: The engineering goal is to bind the abstract physical equations to the actual operating state of the current system. The system will obtain the current operating state parameters of all device nodes in the local network subgraph in batches through the digital configuration system interface. These parameters are real-time data; for example, the inlet pressure of a node is 1.2 MPa, the flow rate is 50 cubic meters per hour, and the operating frequency of a water pump is 48 Hz. The system uses these real-time values ​​as initial conditions, substituting them into the core physical law equations selected in the previous step to assign initial values ​​to the state variables in the equations.

[0056] Step 3: Mathematical Transformation of Topological Relationships and Completion of the Simulation Sandbox: The engineering objective of this step is to transform the physical connections between devices into mathematical constraints between equations, thereby forming a logically complete and solvable system of equations. The system will parse the topological connections of the local network subgraph. For devices connected in series, such as a valve connected to a pump outlet, the system will set the output of the pump outlet pressure equation as the input of the valve inlet pressure equation. For parallel or converging nodes, the system will establish algebraic constraint equations based on the law of conservation of mass, i.e., the sum of all flows into the node must equal the sum of all flows out of the node. Simultaneously, for the boundaries of the subgraph, i.e., the points where the connection to the main network is broken, the system will set their current real-time parameters, such as pressure or flow rate, as constant boundary conditions during the simulation process. By applying initial conditions, boundary conditions, and connection conditions to the core physical law equations, the system completes the initial construction of this local simulation sandbox, making it a complete system of differential-algebraic equations that can be directly used for numerical solutions.

[0057] In a further preferred embodiment, multiple candidate control strategies are simulated in parallel within a local simulation sandbox to generate future state simulation results. This includes: retrieving and generating a set of basic handling actions from a rule base storing handling rules based on the type and anomaly nature of the pressure source equipment; deriving multiple candidate control strategies by adjusting or combining the parameters of the basic handling actions; converting each candidate control strategy into a series of control command sequences with time steps, and using the control command sequences as input to drive the initialized local simulation sandbox to perform numerical solutions, outputting the future state simulation results corresponding to each candidate control strategy.

[0058] Specifically, the engineering objective of this invention in parallel simulating multiple candidate control strategies in a local simulation sandbox is to predict the dynamic response and potential consequences of different response schemes before actually executing any control action through efficient virtual experiments. This aims to scientifically select the optimal solution from multiple possibilities, achieving "foresight" and "optimal selection" in decision-making, thereby avoiding control failures or secondary disasters that may result from blind trial and error or reliance on a single experience. The detailed engineering implementation steps are as follows: Step 1: Generation and diversification of execution strategies: Based on the type and abnormal nature of the pressure source equipment determined in the previous steps, such as "No. 2 refrigeration pump, pressure too high," the system searches in a preset handling rule base. This rule base is a structured knowledge base that stores basic handling actions for common faults of various equipment. The system may match one or more basic handling actions, such as "reduce pump frequency" or "open bypass relief valve." To explore better control effects, the system parameterizes and combines these basic handling actions to generate multiple different candidate control strategy sequences. For example, strategy one is to "reduce the pump frequency to 45 Hz", strategy two is to "reduce the pump frequency to 40 Hz", and strategy three is to "reduce the pump frequency to 45 Hz and simultaneously open the bypass valve by 10%".

[0059] The second step involves discretizing each candidate control strategy sequence in the time dimension. The engineering goal is to transform continuous control intentions into a series of accurate, timestamped instructions executable by the simulation engine. The system sets a total execution and observation duration for each candidate control strategy sequence, for example, 60 seconds. Then, based on the simulation step size, for example, 1 second, the entire strategy is decomposed into a series of control instructions arranged in chronological order. For actions completed instantaneously, such as setting a frequency, it will be placed at the beginning of the time sequence; for actions requiring a process, such as opening a valve, it will be decomposed into instructions for multiple intermediate steps.

[0060] Step 3: Initiate parallel simulation and output results: The system injects the multiple discretized control command sequences generated in the previous step as inputs into the initialized local simulation sandbox. The simulation engine, such as using a numerical integrator like the Runge-Kutta method, will start from the initial state and solve the differential-algebraic equations step by step according to the set simulation step size, simulating the change of all key state parameters in the local network subgraph over time after executing the corresponding control command sequence. This process is performed in parallel for all candidate control strategy sequences. After the simulation is completed, for each candidate strategy, the system outputs a multi-dimensional time series dataset, i.e., the future state projection result. This result records in detail the complete trajectory of the evolution of key parameters such as pressure, flow rate, and temperature of all nodes in the local network subgraph over a preset future time period, such as the next 60 seconds. These parallel projection results will serve as direct inputs for the next step of safety evaluation.

[0061] S5. Quantitatively score the future state projection results according to the multi-dimensional security evaluation rules, select the candidate control strategy with the highest score as the optimal control command, and collect the actual response data after the command is executed.

[0062] In a preferred embodiment, the future state projection results are quantitatively scored according to multi-dimensional safety evaluation rules, and the candidate control strategy with the highest score is selected as the optimal control instruction. This includes: extracting stability indicators, cascading risk indicators, operational energy efficiency indicators, and equipment load indicators from the future state projection results; calculating dimensional scores for the stability indicators, cascading risk indicators, operational energy efficiency indicators, and equipment load indicators respectively; multiplying each dimensional score by its corresponding preset dimensional weight coefficient, and summing all weighted dimensional scores to generate a comprehensive score for each candidate control strategy. The optimal control instruction is the candidate control strategy with the highest comprehensive score.

[0063] Specifically, the engineering objective of this invention, which scores the future state projection results according to preset multi-dimensional safety evaluation rules, is to establish a standardized and quantifiable decision-making evaluation system. This system can objectively and comprehensively measure the combined impact of each candidate control strategy on the future stability, safety, economy, and equipment health of the system while solving the current problem, thus providing a scientific and reliable ranking basis for the final strategy selection. The detailed engineering implementation steps are as follows: First, evaluate the system stability dimension of each future state projection result: This step aims to evaluate whether the system can quickly recover and maintain stable operation after the strategy is implemented. The system extracts the latter half of the time series data of key state parameters (such as the outlet pressure of pressure source equipment) in the projection results and calculates their standard deviation or fluctuation range. If this value is less than the preset stability threshold, the system is considered to have reached stability. Stability dimension score. It is a function calculated based on the rate at which stability is reached and the amplitude of fluctuations at the final steady state. The value of this function is directly proportional to the rate of stability and inversely proportional to the amplitude of fluctuations.

[0064] The second step, assessing the cascading risk dimension, aims to evaluate the global impact of control strategies, particularly identifying and suppressing strategies that, while mitigating the current anomaly, may transmit stress to other vulnerable points through network coupling, potentially triggering a cascading failure. The system examines parameter changes for device nodes directly associated with the subgraph, located at or outside the local network subgraph boundary. If, during the simulation period, the parameters of these nodes exceed their safety thresholds—for example, the pressure in a remote branch falls below the cavitation prevention lower limit—the system records a cascading risk event. (Cascading risk dimension score) It is a penalty function that is calculated based on the number and severity of the triggered chain of risk events. Its initial value is full marks, and it is deducted non-linearly according to the number and severity of the risk events.

[0065] Step 3: Perform an assessment of operational energy efficiency. This step aims to integrate economic considerations into technical decisions and optimize energy-saving solutions. The system will calculate the total energy consumption over the entire simulation period based on the operating parameters of energy-consuming equipment (such as pumps and fans) in the simulation results, including frequency and power, combined with the equipment's energy efficiency curve model. Operational energy efficiency score. It is a mapping function that is monotonically decreasing with the estimated total energy consumption; the lower the energy consumption, the higher the score.

[0066] Step 4: Equipment Load Dimension Assessment: This step aims to ensure the long-term health of equipment and avoid accelerated wear and tear due to improper handling. The system analyzes the amplitude and frequency of actions of key actuators (such as valves and motors) in the simulation results, as well as their final steady-state operating points. If the strategy leads to frequent start-ups and shutdowns, significant adjustments, or prolonged operation in high-load or uneconomical zones, the system will deduct points based on the degree of deviation from the equipment's healthy operating curve. Equipment Load Dimension Score It is a function that comprehensively quantifies the intensity of the equipment's actions and the degree to which the steady-state operating point deviates from the healthy zone. The smoother the actions and the better the operating point, the higher the score.

[0067] Step 5: Perform weighted fusion of the multi-dimensional scores to calculate the final comprehensive score. The system assigns a preset weight coefficient to the score of each of the above dimensions. These weight coefficients are pre-configured by operations and maintenance experts based on the current system's operations and maintenance strategy (such as security priority, energy saving priority), and their sum is 1. The system uses the following formula to calculate the comprehensive score of each candidate control strategy:

[0068] In the formula, This is the final overall score; , , , These are preset weighting coefficients for four dimensions: system stability, cascading risks, operational efficiency, and equipment load. For example, under a safety-first strategy, these are set as follows: , , , ; , , , This corresponds to the score in the respective dimension. The system will calculate a comprehensive score for all candidate control strategies and rank them accordingly.

[0069] S6. Compare the actual response data with the future state projection results to generate policy effect deviation data. Based on this, adjust the weight coefficients in the initial weighted mapping network to generate an updated weighted mapping network for subsequent control cycles.

[0070] In a preferred embodiment, adjusting the weight coefficients in the initial weighted mapping network to generate an updated weighted mapping network for subsequent control cycles includes: analyzing policy effect deviation data; if the deviation data is less than a preset consistency threshold, increasing the weight coefficients of mapping relationships related to policy execution within a local network subgraph according to a preset positive update function; if the deviation data in the policy effect deviation data is greater than a preset mismatch threshold, locating the device with the largest deviation, and decreasing the weight coefficients of mapping relationships directly associated with that device node according to a preset negative update function; and periodically performing a global smoothing adjustment of the weight coefficients of all mapping relationships in the entire network based on the statistical distribution of all historical policy effect deviation data to generate an updated weighted mapping network.

[0071] Specifically, the engineering objective of adaptively adjusting the weight coefficients of the mapping relationship based on strategy effect deviation data in this invention is to construct a dynamic, self-correcting data fusion reliability system. This system, by continuously comparing the model's predictions with the actual responses of the physical world, allows the system to learn and correct its judgment of the data source's reliability. This makes the foundation of the entire intelligent decision-making system—the fusion relationship between BIM static structure and configuration dynamic data—increasingly accurate and reliable over time. The detailed engineering implementation steps are as follows: Step 1: Perform instant weight adjustment based on a single event: After a coordinated control event is completed, the system obtains the strategy effect deviation data as input. This data records in detail the differences between the actual response curves and simulation curves of key parameters. The system analyzes this deviation data. If the root mean square error of the deviation is less than a preset consistency threshold, for example, less than 2% of the predicted value range, the system determines that the simulation model highly matches physical reality. Based on this positive feedback, the system increases the weight coefficients of the mapping relationships corresponding to the pressure root cause devices involved in this event and all devices within their respective local network subgraphs. Conversely, if the root mean square error exceeds a preset deviation threshold, the system will locate the specific device or parameter point contributing the largest deviation and lower the weight coefficient of its corresponding mapping relationship. Simultaneously, the system will mark this mapping relationship as unobservable and automatically trigger a deep verification process based on physical consistency rules to investigate potential sensor malfunctions or configuration errors.

[0072] The second step involves performing periodic smoothing updates based on long-term statistics. The goal is to avoid excessive disturbances to the weighting system caused by the randomness of a single event, thereby improving the long-term stability and robustness of the weight coefficients. The system maintains a database in the background containing historical strategy performance deviation data. At a preset, relatively long period, such as every 24 hours, the system initiates a global weight smoothing update task. This task statistically analyzes the average deviation performance of each mapping relationship over the past period. For mappings that consistently exhibit low bias over a long period, the system assigns a small gain to their weight coefficients; while for mappings that repeatedly show large bias, a continuous and gradual weight decay is applied. This smooth update process uses the following formula:

[0073] In the formula, These are the updated weighting coefficients; These are the weighting coefficients before the update; It is the average normalized deviation of this mapping relationship over the past period; It is a learning rate factor used to control the impact of single-cycle bias on weight updates. It is preset by algorithm engineers based on experience or through experimental optimization. For example... (Slow learning rate, suitable for stable systems with high noise) (Fast learning rate, suitable for systems that need to respond quickly to changes).

[0074] Finally, by combining the aforementioned dual mechanisms of real-time adjustment and periodic smooth updates, the system achieves dynamic and adaptive management of the weight coefficients of all mapping relationships. This ensures that in subsequent virtual stress calculations and transmission processes, the system can automatically assign higher trust to data sources that have proven to be more reliable historically, thereby continuously improving the decision-making accuracy and reliability of the entire linkage control system.

[0075] In a preferred embodiment, the method further includes: identifying systematic deviations related to physical processes in the strategy effect deviation data and locating the core physical law equations that cause the deviations in the local simulation sandbox; using actual response data as the target, solving the key model parameters in the core physical law equations in reverse using a parameter identification algorithm; and weighting and fusing the key model parameters solved in reverse with the original parameters in the model library to generate an updated physical model library.

[0076] Specifically, the engineering objective of adaptively adjusting model parameters in the simplified physical model library based on strategy effect deviation data in this invention is to transform the simulation model from a "static knowledge base" to a "dynamic evolutionary entity." By continuously utilizing real-world physical feedback to calibrate and optimize key parameters within the model, the system can ensure that the local simulation sandbox used for deduction becomes increasingly faithful to physical reality over time, thereby improving the predictability and accuracy of automated decision-making. The detailed engineering implementation steps are as follows: Step 1: Deviation Source Tracing and Model Localization: After a coordinated control event is completed, the system receives strategy effect deviation data as input. The system will conduct in-depth analysis of this data, focusing not only on the magnitude of the deviation but, more importantly, on identifying the pattern of deviation generation. For example, the system might find that the actual pressure response is 2 seconds slower than the simulation deduction, or the steady-state flow rate is 5% lower than the deduction result. Based on the characteristics of these deviations and the composition of the local simulation sandbox involved in this event, the system will locate the physical element most likely causing the deviation and its corresponding model equation. For example, the deviation in response delay is likely related to the inertial time constant of the pump model, while the deviation in steady-state flow rate may point to the friction coefficient of the pipe model.

[0077] The third step is to perform parameter fusion and update of the model library: After the parameter identification algorithm converges, it will output one or a set of inverted key parameters that are optimal in this event. To ensure the stability and generalization ability of the model library, the system will not directly overwrite the original parameters in the model library with this new parameter. Instead, a weighted fusion strategy is used for updating. The update formula is:

[0078] In the formula, These are the updated model parameters; These are parameters that already exist in the model library; These are the optimal parameters obtained from the inversion of this event; This is an update weight, or learning rate, typically with a small value, such as 0.05 to 0.2, used to control the magnitude of correction to long-term model parameters by a single event. Through this gradual fusion update, the pre-defined simplified physics model library can continuously absorb experience from each real-world interaction, making its internal model parameters increasingly closer to the real physical characteristics of the managed facilities, thus providing a more solid and accurate foundation for each future simulation.

[0079] The above content is merely an example and illustration of the concept of the present invention. Those skilled in the art can make various modifications or additions to the specific embodiments described, or use similar methods to replace them, as long as they do not deviate from the concept of the invention or exceed the scope defined by the present invention, and all such modifications and additions should fall within the protection scope of the present invention.

Claims

1. A method for linking digital configuration and BIM mapping of prefabricated electromechanical facilities, characterized in that, include: S1. Obtain the physical connection relationships of equipment in the building information model and the real-time monitoring data in the digital configuration system, and establish an initial correspondence based on the unique identifier or tag number of the equipment to generate an initial weighted mapping network; S2. Based on the initial weighted mapping network, calculate the deviation of the real-time monitoring data of each monitoring point from its target value range, generate the initial virtual pressure value of each node, and use the initial weighted mapping network to transmit and superimpose the pressure to generate a dynamic pressure distribution map representing the global risk situation; S3. Identify the abnormal pressure nodes in the dynamic pressure distribution map, and perform reverse pressure gradient tracing to determine the pressure root cause equipment, thereby extracting a local network subgraph containing the pressure root cause equipment and its affected neighbor equipment; S4. Call the physical model equations that match the equipment type in the local network subgraph, and combine the topological connection relationships of the local network subgraph to construct a local simulation sandbox, and deduce multiple candidate control strategies in parallel in the local simulation sandbox to generate future state deduction results; S5. Quantitatively score the future state projection results according to the multi-dimensional safety evaluation rules, select the candidate control strategy with the highest score as the optimal control instruction, and collect the actual response data after the instruction is executed; S6. Compare the actual response data with the future state projection results to generate strategy effect deviation data, and adjust the weight coefficients in the initial weighted mapping network accordingly to generate the updated weighted mapping network for subsequent control cycles.

2. The method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping according to claim 1, characterized in that, The initial weighted mapping network is generated by: parsing the 3D data structure of the building information model, extracting the physical connection relationships between electromechanical equipment nodes and equipment endpoints, and constructing an equipment physical topology map reflecting the spatial layout of the facility; traversing the monitoring point list in the digital configuration system, and establishing dynamic associations between real-time data streams and nodes in the equipment physical topology map based on the corresponding rules of equipment unique identifiers or tag numbers; extracting physical attribute parameters of physical connection relationships, quantifying the transmission influence coefficient of each connection edge according to the physical attribute parameters, and assigning this as a weight value to the corresponding edge of the equipment physical topology map to form the initial weighted mapping network.

3. The method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping according to claim 1, characterized in that, The initial virtual pressure value for each node is generated by: obtaining the real-time monitoring data corresponding to each monitoring point and its target value range stored in the device attribute library; calculating the normalized value of the real-time monitoring data deviating from the target value range to obtain the deviation degree; obtaining the device criticality weight associated with the monitoring point and the duration of the current abnormal state; and calculating the initial virtual pressure value by positively weighting the deviation degree and the device criticality weight, and then performing logarithmic processing on the duration of the current abnormal state.

4. The method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping according to claim 2, characterized in that, The pressure is transmitted and superimposed using an initial weighted mapping network to generate a dynamic pressure distribution map representing the global risk situation. This includes: obtaining the transmission influence coefficient assigned to each connection edge in the initial weighted mapping network; multiplying the comprehensive virtual pressure value of the upstream node by the transmission influence coefficient of the connection edge to obtain the attenuated pressure value transmitted to the downstream node; at each device node, summing the node's own initial virtual pressure value with all attenuated pressure values ​​transmitted from the upstream nodes to generate a new round of comprehensive virtual pressure value for the node. The comprehensive virtual pressure values ​​of all nodes constitute the dynamic pressure distribution map.

5. The method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping according to claim 1, characterized in that, Identifying abnormal pressure nodes in a dynamic pressure distribution map and performing reverse pressure gradient tracing to determine the root cause device of the pressure includes: identifying abnormal pressure nodes in the dynamic pressure distribution map whose comprehensive virtual pressure value exceeds a preset alarm threshold; calculating the virtual pressure difference between the abnormal pressure node and all directly connected upstream nodes, generating a pressure change gradient set; selecting the connection path with the largest gradient value from the pressure change gradient set as the main conduction path; setting the upstream node of the main conduction path as the new current node, and repeating the steps of calculating the pressure change gradient and selecting the main conduction path until the virtual pressure value of the current node is lower than the preset root cause determination threshold, and finally determining the node that meets the conditions as the pressure root cause device.

6. The method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping according to claim 1, characterized in that, Constructing a local simulation sandbox includes: matching each device in the local network subgraph with a corresponding core physical law equation from a model library that stores physical models of various devices; obtaining the real-time operating parameters of each device in the local network subgraph as the initial conditions for the core physical law equations; and transforming the topological connection relationship of the local network subgraph into boundary constraints and flow conservation constraints between the core physical law equations to complete the initialization of the local simulation sandbox.

7. The method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping according to claim 1, characterized in that, Multiple candidate control strategies are simulated in parallel within a local simulation sandbox to generate future state simulation results. This includes: retrieving and generating a set of basic handling actions from a rule base storing handling rules based on the type and anomaly nature of the pressure source equipment; deriving multiple candidate control strategies by adjusting or combining the parameters of the basic handling actions; converting each candidate control strategy into a series of control command sequences with time steps, and using the control command sequences as input to drive the initialized local simulation sandbox for numerical solution, outputting the future state simulation results corresponding to each candidate control strategy.

8. The method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping according to claim 1, characterized in that, The future state simulation results are quantitatively scored according to multi-dimensional safety evaluation rules. The candidate control strategy with the highest score is selected as the optimal control instruction. This includes: extracting stability indicators, cascading risk indicators, operational energy efficiency indicators, and equipment load indicators from the future state simulation results; calculating dimensional scores for stability indicators, cascading risk indicators, operational energy efficiency indicators, and equipment load indicators respectively; multiplying each dimensional score by its corresponding preset dimensional weight coefficient, and summing all weighted dimensional scores to generate a comprehensive score for each candidate control strategy. The optimal control instruction is the candidate control strategy with the highest comprehensive score.

9. The method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping according to claim 1, characterized in that, Adjusting the weight coefficients in the initial weighted mapping network to generate an updated weighted mapping network for subsequent control cycles includes: analyzing policy effect deviation data; if the deviation data is less than a preset consistency threshold, increasing the weight coefficients of mapping relationships related to policy execution within the local network subgraph according to a preset positive update function; if the deviation data in the policy effect deviation data is greater than a preset mismatch threshold, locating the device with the largest deviation, and decreasing the weight coefficients of mapping relationships directly associated with that device node according to a preset negative update function; periodically performing global smoothing adjustments on the weight coefficients of all mapping relationships across the entire network based on the statistical distribution of all historical policy effect deviation data, thereby generating an updated weighted mapping network.

10. The method for linking prefabricated electromechanical facilities with digital configuration and BIM mapping according to claim 1, characterized in that, Also includes: Identify systematic deviations related to physical processes in the strategy effect deviation data and locate the core physical law equations that cause these deviations in the local simulation sandbox; Using actual response data as the target, the key model parameters in the core physical law equations are solved in reverse through parameter identification algorithms; the key model parameters solved in reverse are then weighted and fused with the original parameters in the model library to generate an updated physical model library.

Images