A cross-regional building decoration engineering data collaborative analysis system

Abstract

The application relates to the technical field of building information data processing, and discloses a cross-regional building decoration engineering data collaborative analysis system which comprises a coordination server and multiple edge nodes. The application calculates the topological out-degree density truncated full-amount state data message of a first influence subgraph through the edge nodes, transmits an incremental operation vector and a truncated feature code to a data communication link, and converts a state change amount value into a fixed-point number format in a node local closed loop to perform calculation again. The coordination server applies an independent memory address segment to construct an independent sandbox and generate a shadow subgraph, a topological degradation constraint solving engine executes step attenuation calculation to find a legal compensation parameter in combination with a rigid constraint boundary condition, an edge node receives a compensation operation vector carrying an expected version number and executes an overwrite operation by using an optimistic lock mechanism, and the coordination server triggers a calculation depth fuse mechanism to forcibly terminate a calculation process.

Description

Technical Field

[0001] This invention relates to the field of building information data processing technology, specifically a collaborative analysis system for cross-regional building decoration engineering data. Background Technology

[0002] Data synchronization for cross-regional architectural decoration projects faces wide area network bandwidth limitations. Network serialization and packetization operations transform affected graph nodes into full-state data packets, which pose a risk of bandwidth exhaustion during cross-regional network transmission. Furthermore, servers deployed across regions exhibit heterogeneous CPUs. These heterogeneous CPUs rely on floating-point rounding rules to process state change values, and differences in underlying hardware calculation rules lead to errors in underlying data coordination and state calculation.

[0003] Modifications to engineering data occurring in different regions send commands to the central computer room, triggering concurrent modifications across multiple nodes. These modifications, targeting the same decorative project entity dimensions, lead to spatial parameter interference conflicts. The central computer room's complex interference compensation calculations directly consume the main database's runtime environment, and the spatial parameter compensation derivation process disrupts the main database's normal read / write response mechanism.

[0004] Concurrent modification operations submitted to the underlying database for overwrite writes rely on a mutual exclusion lock acquisition mechanism. The reliance on mutual exclusion locks for underlying database state synchronization can easily lead to resource deadlocks. Faced with irreconcilable spatial conflicts, the background inference engine searches for compensation parameters, performing multi-dimensional optimization exploration that falls into an infinite loop. This unsolvable infinite loop operation causes excessive consumption of server memory resources.

[0005] Therefore, this invention proposes a collaborative analysis system for cross-regional architectural decoration engineering data to address the shortcomings of existing technologies. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this invention provides a collaborative analysis system for cross-regional architectural decoration engineering data. It solves the following problems: cross-regional server execution of engineering data synchronization encounters wide area network bandwidth limitations leading to obstruction of full-scale status data message transmission; heterogeneous central processing units cause floating-point calculation errors due to differences in underlying hardware calculation rules; concurrent modifications by multiple nodes cause spatial parameter interference conflicts and interference compensation calculations interfere with the normal read / write response mechanism of the main database; the mutual exclusion lock application process of the underlying database causes resource deadlock; and the inference engine gets stuck in an unsolvable infinite loop, causing server memory resource consumption.

[0007] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a collaborative analysis system for cross-regional building decoration engineering data, including a coordination server deployed in a central computer room and multiple edge nodes deployed in different regions. The multiple edge nodes are divided into a first edge node deployed in a first region and a second edge node deployed in a second region.

[0008] The collaborative analysis system reads global architectural decoration engineering data and separates it into different decoration engineering entity data. The collaborative analysis system abstracts the different decoration engineering entity data into independent vertices. The collaborative analysis system generates directed edges connecting independent vertices according to spatial support and subordinate relationships. The collaborative analysis system performs topological association and assembly of all independent vertices and directed edges connecting independent vertices to generate a global engineering-dependent directed acyclic graph. The collaborative analysis system assigns topological depth attribute values ​​to the independent vertices in the global engineering-dependent directed acyclic graph according to the spatial arrangement order.

[0009] The first edge node intercepts local engineering data modification actions and extracts the state change value. The first edge node integrates a low-level formatting module, which converts the state change value into a fixed-point number format. The first edge node reads the node identifier of the independent vertex that has undergone modification. The first edge node extracts the monotonically increasing sequence to generate a logical clock stamp value. The first edge node concatenates and encapsulates the node identifier, the state change value converted to a fixed-point number format, the current topology depth attribute value, and the logical clock stamp value to generate an incremental operation vector.

[0010] The first edge node uses the independent vertex corresponding to the node identifier as the starting vertex for searching and searches for successor nodes layer by layer within the globally dependent directed acyclic graph. The first edge node merges the starting vertex and all found successor nodes and separates them to generate the first influence subgraph. The first edge node calculates the topological out-degree density of the first influence subgraph. When the topological out-degree density is greater than the network serialization cost threshold, the first edge node forcibly truncates the first influence subgraph and packages it into a full state data packet, generating a truncation feature code. The first edge node sends the incremental operation vector and the truncation feature code to the second edge node.

[0011] The second edge node receives the incremental operation vector and the truncated feature code. The second edge node integrates a fixed-point number arithmetic engine. Based on the truncated feature code, the second edge node triggers the call of the integrated fixed-point number arithmetic engine to retrieve and query the independent vertex corresponding to the node identifier in the globally dependent directed acyclic graph that is synchronized in local memory as the starting point for calculation. The second edge node calls the fixed-point number arithmetic engine to calculate the state of the first influence subgraph by locally closing the loop in the second edge node according to the derivation rules.

[0012] The second edge node extracts the range parameters affected by the local engineering data modification actions that have not yet been merged and written back to the underlying database, and encapsulates them with the logical clock stamp values ​​generated locally based on the global time synchronization protocol to generate a second uncommitted impact subgraph. The second edge node extracts the node identifiers corresponding to the independent vertices covered by the state of the first impact subgraph to generate a first vertex identifier set, and the second edge node extracts the node identifiers corresponding to the independent vertices covered by the second uncommitted impact subgraph to generate a second vertex identifier set.

[0013] The second edge node performs a memory data overlap calculation operation on the first vertex identifier set and the second vertex identifier set to obtain the intersection detection operation result. When the intersection detection operation result produces an intersection conflict, the second edge node extracts the state of the first influencing subgraph and the second uncommitted influencing subgraph to form conflict concurrent data carrying their respective corresponding logical clock stamp values. The second edge node sends the conflict concurrent data to the coordination server.

[0014] The coordination server receives conflicting concurrent data and starts a topology degradation constraint solving engine within an independent sandbox to find a safe building parameter range. The coordination server generates a compensation operation vector based on the safe building parameter range and broadcasts it to multiple edge nodes.

[0015] The coordination server parses and extracts the conflicting concurrent data, including the conflicting operations. The coordination server then multiplies the topology depth attribute value and the corresponding logical clock stamp value of the conflicting operation by a preset topology depth weight allocation coefficient and a time series weight allocation coefficient, and adds them together to obtain the priority value.

[0016] The coordinating server classifies operations with higher priority values ​​as winning operations and uniformly classifies operations with lower priority values ​​as losing operations. The coordinating server requests and allocates an independent memory address segment as an isolated memory space. Within this isolated memory space, the coordinating server constructs an independent sandbox. Within this sandbox, the coordinating server generates a shadow subgraph from the data structure copies created by deep copying the decoration project entity nodes associated with the losing operations from the main database.

[0017] The coordination server converts the engineering state data corresponding to the winning operation into mathematical inequality constraints, setting these as rigid constraint boundary conditions. The coordination server then passes these rigid constraint boundary conditions to the shadow subgraph and launches the topology degradation constraint solving engine within an independent sandbox. The topology degradation constraint solving engine reads the original modification difference that the losing operation was originally expected to write from the shadow subgraph as the initial iteration reference point for parameter exploration. It then performs step-size decay calculations based on the rigid constraint boundary conditions to find the safe building parameter range. Finally, the topology degradation constraint solving engine establishes the compensation modification difference generated by the step-size decay calculation as the final valid compensation parameter.

[0018] The coordination server concatenates and reassembles the valid compensation parameters, the corresponding node identifier, and the latest logical clock value corresponding to the successful operation into a compensation operation vector carrying the expected version number. Multiple edge nodes receive the compensation operation vector and parse the valid compensation parameters and associated expected version number. Multiple edge nodes use an optimistic locking mechanism to perform concurrent verification and judgment calculations. When the concurrent verification and judgment calculations pass, multiple edge nodes submit the data to the underlying database for overwrite writing.

[0019] The topology degradation constraint solving engine increments the iteration counter each time it performs step decay calculation. The coordination server reads the current iteration calculation depth order generated by the iteration counter in real time and compares the current iteration calculation depth order with the maximum iteration depth threshold.

[0020] When the current iteration calculation depth is greater than or equal to the maximum iteration depth threshold and no valid compensation parameters have been established, the coordination server triggers the calculation depth circuit breaker mechanism to forcibly terminate the subsequent calculation process of the topology degradation constraint solving engine. The coordination server generates a manual intervention alarm message containing detailed conflict context information and sends it to the source edge node to force a manual offline negotiation and correction process.

[0021] The technical solution of this invention clarifies the collaborative analysis process for cross-regional architectural decoration engineering data.

[0022] The collaborative analysis system extracts node connection dependencies to generate a global engineering dependency directed acyclic graph (DAG). It then uses the first edge node to peel off the first influence subgraph and calculates the topology out-degree density. When the topology out-degree density exceeds the network serialization cost threshold, the collaborative analysis system intercepts the outgoing delivery of all state data packets. The first edge node calls the underlying formatting module to convert the state change values ​​(containing decimal information) into a fixed-point number format that records the change differences using pure integer encoding. The WAN bidirectional data communication link only transmits the incremental operation vector and truncated feature code. The second edge node calls its locally integrated fixed-point number arithmetic engine to perform closed-loop replay calculations. The fixed-point number format eliminates computational errors caused by differences in underlying hardware calculation rules among heterogeneous CPUs, and the out-degree density truncation avoids the risk of bandwidth exhaustion caused by cross-regional network transmission in a WAN environment.

[0023] The second edge node performs an intersection detection operation on the second unsubmitted impact subgraph and the state of the first impact subgraph generated by the closed-loop replay calculation. When the intersection detection operation results in an intersection conflict, the collaborative analysis system initiates an arbitration compensation calculation process by the coordination server. The coordination server determines the priority value by combining the topology depth attribute value and the logical clock stamp value to allocate a weight product. The coordination server requests and allocates an independent memory address segment not shared with the main business process to build an independent sandbox and generates a shadow subgraph within the independent sandbox.

[0024] The coordination server transforms the engineering state data corresponding to the winning operation into rigid constraint boundary conditions and inputs them into the shadow subgraph. The topology degradation constraint solving engine reads the original modification difference that the losing operation is expected to write within the shadow subgraph as the initial iteration benchmark for parameter exploration and performs step size decay calculation.

[0025] The coordination server identifies the compensation modification difference falling within the feasible range defined by rigid constraint boundary conditions as a valid compensation parameter. The coordination server then concatenates the expected version number and broadcasts the compensation operation vector to multiple edge nodes. These edge nodes then utilize an optimistic locking mechanism to perform concurrent verification and calculation to complete the underlying database overwrite operation.

[0026] The independent sandbox, at the physical memory level, blocks interference from complex compensation calculations on the main database's normal read / write response mechanism. Step-size decay calculations find safe intervals for building parameters to resolve spatial parameter interference conflicts. The coordination server uses an iteration counter to monitor the number of iterations in the parameter exploration loop. When the current iteration depth is greater than or equal to the maximum iteration depth threshold, a computation depth circuit breaker mechanism is triggered, destroying the independent sandbox environment and returning the independent memory address segment. This avoids infinite loops in unsolvable states and excessive memory resource consumption.

[0027] This invention provides a collaborative analysis system for cross-regional architectural decoration engineering data. It has the following beneficial effects: 1. This invention calculates the topology out-degree density of the first influence subgraph through a first edge node. When the topology out-degree density exceeds the network serialization cost threshold, the full state data packet is truncated, and an incremental operation vector and truncated feature code are sent to a second edge node. The second edge node then uses its internally integrated fixed-point arithmetic engine to locally recalculate the state of the first influence subgraph. The collaborative analysis system transmits the incremental operation vector and truncated feature code only through the bidirectional data communication link, avoiding the risk of bandwidth exhaustion caused by cross-regional network transmission in wide area networks. The underlying formatting module converts the state change values ​​into a fixed-point format, which eliminates calculation errors caused by differences in underlying hardware calculation rules among heterogeneous central processing units.

[0028] 2. This invention allocates independent memory address segments through a coordination server to define isolated memory spaces and constructs independent sandboxes within these spaces. Within these sandboxes, the coordination server generates shadow subgraphs from the data structure copies created by deep copying. The topology degradation constraint solving engine then performs step-size decay calculations within these shadow subgraphs, incorporating rigid constraint boundary conditions, to find safe building parameter ranges. The independent sandbox, at the physical memory level, prevents interference from complex compensation calculations on the normal read / write response mechanism of the main database. The topology degradation constraint solving engine uses the engineering state data corresponding to successful operations as a constraint range to explore parameters for failed operations, searching for legal compensation parameters that fall within the feasible range defined by rigid constraint boundary conditions, thus resolving spatial parameter interference conflicts caused by engineering data modification actions.

[0029] 3. This invention uses a coordinating server to concatenate the desired version number and broadcast a compensation operation vector. Edge nodes receive the compensation operation vector and perform concurrent verification calculations using an optimistic locking mechanism. When the concurrent verification calculation passes, the node submits the calculation to the underlying database for overwrite. The coordinating server triggers a computational depth circuit breaker mechanism when the current iteration computation depth is greater than or equal to the maximum iteration depth threshold. Edge nodes use the optimistic locking mechanism and the desired version number to synchronize the local underlying database state, avoiding the risk of resource deadlocks that can easily occur during the mutual exclusion lock acquisition process. The computational depth circuit breaker mechanism forcibly terminates the subsequent computation process of the topology degradation constraint solving engine and returns the independent memory address segment to the underlying operating system, preventing infinite loop operations in the unsolvable state and excessive consumption of memory resources. Attached Figure Description

[0030] Figure 1 This is a system block diagram of the present invention.

[0031] Figure 2 This is a flowchart of the method of the present invention.

[0032] Figure 3 This diagram illustrates the parameter optimization process of the topology degradation constraint solving engine of this invention. Detailed Implementation

[0033] 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.

[0034] See attached document Figure 1 This invention provides a collaborative analysis system for cross-regional architectural decoration engineering data, comprising: The coordination server is deployed in the central computer room and multiple edge nodes are deployed in different regions. The multiple edge nodes and the coordination server establish a two-way data communication link through the wide area network. The multiple edge nodes are divided into a first edge node deployed in the first region and a second edge node deployed in the second region. The first edge node and the second edge node independently intercept local engineering data modification actions and perform state calculations. The second edge node receives communication data sent by the first edge node and performs closed-loop replay calculations. The coordination server receives conflicting concurrent data and performs arbitration compensation calculations.

[0035] The first edge node integrates a low-level formatting module and a fixed-point number arithmetic engine. The low-level formatting module intercepts write instructions and converts the state change values ​​into fixed-point number format. The fixed-point number arithmetic engine eliminates calculation errors caused by differences in calculation rules between heterogeneous central processing units in different regions. The first edge node peels off the first influence subgraph based on the topological depth attribute value of the directed acyclic graph that the global project depends on. The first edge node truncates the full state data message and sends the incremental operation vector and truncation feature code to the second edge node.

[0036] The second edge node calls the fixed-point arithmetic engine integrated within the second edge node to calculate the state of the first influence subgraph based on the deterministic derivation rules and closed-loop replay. The second edge node performs an intersection detection operation on the state of the first influence subgraph and the second uncommitted influence subgraph and performs a lock-free merging operation. When the intersection detection operation results in an intersection conflict, the coordination server calculates the priority values ​​for the multiple operations that are in conflict. The coordination server delineates an independent sandbox in the isolated memory space and generates a shadow subgraph. Inside the independent sandbox, the coordination server starts the topology degradation constraint solving engine to find the feasible building parameter range. The coordination server broadcasts the compensation operation vector to multiple edge nodes to complete the data collaborative processing flow.

[0037] See attached document Figure 2 , Figure 2 This is a flowchart of a collaborative analysis method for cross-regional architectural decoration engineering data according to an embodiment of the present invention. The present invention provides a collaborative analysis method for cross-regional architectural decoration engineering data, comprising the following steps: S10, the collaborative analysis system reads the global building decoration project data, abstracts the different decoration project entity data into independent vertices, generates directed edges connecting independent vertices according to the spatial support and subordinate relationship between decoration project entities, constructs a global project dependency directed acyclic graph, and assigns topological depth attribute values ​​to the independent vertices in the global project dependency directed acyclic graph according to the spatial arrangement order.

[0038] S20, the first edge node intercepts the local engineering data modification action and extracts the state change value. The first edge node calls the underlying formatting module to convert the state change value into a fixed-point number format. The first edge node generates an incremental operation vector with node identifier, state change value and topology depth attribute value. The first edge node traverses the directed acyclic graph with reachable paths downward based on the global engineering dependency. The first edge node peels out and generates the first influence subgraph.

[0039] S30, the first edge node calculates the topology out-degree density of the first influence subgraph. When the topology out-degree density is greater than the network serialization cost threshold, the first edge node truncates the full state data message. The first edge node sends the incremental operation vector and truncation feature code to the second edge node. The second edge node calls the fixed-point number operation engine integrated inside the second edge node to calculate the state of the first influence subgraph by local closed-loop replay according to the derivation rules of the global engineering dependent directed acyclic graph.

[0040] S40, the second edge node performs an intersection detection operation on the state of the first influence subgraph and the second uncommitted influence subgraph. When the intersection detection operation result is an empty set, the second edge node performs a lock-free merge operation and saves it to disk. When the intersection detection operation result is not an empty set, the coordination server calculates the priority value for the conflicting operations.

[0041] S50, the coordination server determines the operation corresponding to the high priority value as a winning operation and executes it to overwrite the database. For the losing operation, the coordination server defines an independent sandbox in the isolated memory space and generates a shadow subgraph. The coordination server passes the engineering state data corresponding to the winning operation as a rigid constraint boundary condition into the shadow subgraph. The coordination server starts the topology degradation constraint solving engine inside the independent sandbox to find the building parameter range. The coordination server converts the building parameter range into a compensation operation vector. The coordination server broadcasts the compensation operation vector to the first edge node and the second edge node.

[0042] The technical solution for the collaborative analysis method of cross-regional architectural decoration engineering data will be explained in detail below with reference to the accompanying drawings and specific working steps.

[0043] Step S10 includes the following sub-steps: S101, the collaborative analysis system reads global building decoration engineering data from the file storage terminal. The global building decoration engineering data records the attribute parameters of various engineering elements within the building environment. The collaborative analysis system calls the parsing module to process the global building decoration engineering data and separate it into different decoration engineering entity data. The decoration engineering entity data specifically refers to the load-bearing wall thickness parameters, ceiling joist size parameters, and electrical pipeline layout coordinate data recorded in the building construction documents.

[0044] S102, the collaborative analysis system abstracts different decoration project entity data into independent vertices, extracts the type codes and numerical parameters of different decoration project entity data to construct a basic data structure, and instantiates the basic data structure into independent vertices in the internal graph database cache space.

[0045] S103, the collaborative analysis system analyzes the spatial support subordinate relationship between different decoration project entity data. The spatial support subordinate relationship reflects the inherent mechanical support state and spatial interference boundary state between the load-bearing bottom structural material and the attached surface material in the spatial environment.

[0046] For the process of parsing and extracting spatial support subordinate relationships from the internal structure information model (BIM) file, those skilled in the art can call basic data parsing tools to perform attribute field stripping. Basic model data graph conversion technology is a well-known technology in this field and will not be described in detail here.

[0047] S104, the collaborative analysis system generates directed edges connecting independent vertices according to the spatial support subordinate relationship. The collaborative analysis system sets the pointing rule of the directed edges to point unidirectionally from the independent vertex corresponding to the load-bearing bottom structural material to the independent vertex corresponding to the attached surface material. The collaborative analysis system takes the independent vertex corresponding to the wall keel as the generation starting point and the independent vertex corresponding to the outer moisture-proof board as the generation ending point. The collaborative analysis system establishes a dependency pointer between the generation starting point and the generation ending point to form a directed edge.

[0048] S105, the collaborative analysis system performs topological association and assembly of all independent vertices and directed edges connecting independent vertices instantiated in the internal graph database cache space. The collaborative analysis system constructs a global engineering dependency directed acyclic graph, which reflects the corresponding association status of all decorative engineering entity components within the global building space. S106, the collaborative analysis system traverses the global project's dependent directed acyclic graph to identify the set of vertices with an in-degree of zero. To prevent isolated decorative elements without connections from causing depth calculation errors, the collaborative analysis system identifies and removes independent non-stressed primitives (such as independently placed soft furnishing plant models) from the set of vertices with an in-degree of zero based on the type codes extracted from the basic data structure. The collaborative analysis system sets the remaining set of vertices with an in-degree of zero after filtering as the reference support vertices corresponding to the underlying load-bearing foundation structure, and assigns the initial value of zero to the topological depth attribute value of the reference support vertices.

[0049] S107, the collaborative analysis system performs layer-by-layer pathfinding calculations to descendant vertices according to the spatial arrangement order from the bottom load-bearing foundation structure to the surface decorative materials.

[0050] For the calculation process of graph topology hierarchical pathfinding, those skilled in the art can call the breadth-first search algorithm to find the longest reachable path length from the baseline supporting vertex to any independent vertex. Graph theory path optimization calculation technology is a well-known technology in this field and will not be elaborated here.

[0051] S108, the collaborative analysis system directly uses the calculated longest reachable path length as the topological depth attribute value of the corresponding independent vertex. The topological depth attribute value represents the vertical hierarchy position of the independent vertex within the global engineering-dependent directed acyclic graph. The collaborative analysis system writes and stores the topological depth attribute value as a basic data field within the basic data structure of the independent vertex.

[0052] Step S20 includes the following sub-steps: S201, the first edge node intercepts the write operation instruction that submits a data modification request to the database at the application software interface. The first edge node parses the data payload message attached to the write operation instruction and extracts the state change value from the data payload message. The state change value represents the variation difference executed by the engineering designer for the specific decoration project entity size.

[0053] S202. Regarding the interception of the execution control process of write operation instructions at the operating system level, those skilled in the art can call the system-level hook function to intercept communication ports. Communication port data packet interception and payload message parsing technology are well-known technologies in this field and will not be described in detail here.

[0054] S203, the first edge node transmits the state change value to the internally integrated low-level formatting module. There is a difference in instruction sets between the heterogeneous central processing units at both ends of the wide area network server terminals. The conventional floating-point arithmetic mechanism has the risk of rounding error. The low-level formatting module forcibly converts the state change value with decimal information to a fixed-point format by arithmetic shifting. The fixed-point format uses pure integer encoding to record the change difference. The fixed-point format avoids the problem of inconsistent floating-point calculation caused by heterogeneous central processing unit environments.

[0055] S204, the first edge node reads the node identifier of the node that underwent modification, extracts the state change value after fixed-point number formatting, and queries the current topology depth attribute value of the node that underwent modification from the global project dependency directed acyclic graph. At the same time, the first edge node calls the local system kernel that maintains global time synchronization with the coordination server to extract a monotonically increasing sequence to generate a logical clock stamp value. The first edge node concatenates the node identifier, state change value, current topology depth attribute value, and logical clock stamp value into memory data segments, encapsulates them to generate an independent data structure object, and outputs it as an incremental operation vector.

[0056] S205, the first edge node takes the node where the modification operation occurred as the starting vertex of the search. The first edge node locates the lower-level dependent directed edges connected to the starting vertex of the search within the global project dependent directed acyclic graph. The first edge node searches down layer by layer along the direction of the lower-level dependent directed edges to find all successor nodes with reachable paths.

[0057] S206. For the process of searching for nodes downwards in a directed acyclic graph data structure, those skilled in the art can call a depth-first search algorithm or a breadth-first search algorithm to access adjacent nodes in the direction of pointer reference. Graph data node traversal and search technology is a well-known technology in this field and will not be described in detail here.

[0058] S207, the first edge node merges the starting vertex of the search and all the found successor nodes into an independent vertex set. The first edge node classifies the directed edges connecting the nodes inside the independent vertex set into a dependent edge set. The first edge node separates the independent vertex set and the dependent edge set from the global project dependent directed acyclic graph to generate an independent data copy. The first edge node sets the independent data copy as the first influence subgraph. The first influence subgraph represents the set of all subsequent project entity primitives affected by the project data modification action in the spatial support subordinate relationship range.

[0059] Step S30 includes the following sub-steps: S301, the first edge node scans the internal data structure topology of the first influence subgraph, the first edge node classifies and counts the number of data entity elements inside the first influence subgraph, and the first edge node records the total number of independent vertices and the total number of dependent edges connecting the independent vertices.

[0060] S302, the first edge node divides the total number of dependent edges by the total number of independent vertices to obtain the corresponding ratio. The first edge node uses this ratio as the topological out-degree density of the first influential subgraph. The first edge node specifically performs the following calculation formula to obtain the topological out-degree density: ; In the formula, The topological out-degree density represents the first-influence subgraph; This represents the total number of dependent edges in the first-influence subgraph; This represents the total number of independent vertices in the first-influence subgraph.

[0061] S303, the first edge node reads the preset network serialization cost threshold from its internal memory. The network serialization cost threshold represents the maximum boundary reference value that allows direct transmission of complex relational graph structure data in a wide area network environment. The first edge node performs a size comparison operation between the calculated topology out-degree density and the network serialization cost threshold.

[0062] S304, when the topology out-degree density is greater than the network serialization cost threshold, the first edge node determines that directly transmitting the current complex graph structure data across regional networks will cause the risk of bandwidth exhaustion. The first edge node forcibly truncates and packages the first affected subgraph into a full state data packet.

[0063] When the topology out-degree density is less than or equal to the network serialization cost threshold, the first edge node determines that the current graph structure data has not exceeded the WAN bandwidth carrying limit. The first edge node does not perform a truncation operation, but directly serializes the first influential subgraph into a full state data packet and delivers it to the sending queue.

[0064] S305. For the serialization of graph structure data in memory and the forwarding of network data packets, those skilled in the art can call the underlying interface of the communication protocol stack to perform byte stream conversion and mapping. Data memory serialization and network layer packet delivery technology are well-known technologies in this field and will not be described in detail here. S306, after truncating the entire state data packet, the first edge node retains only the core operation parameters. The first edge node sets specific bit flags (e.g., setting the protocol header flag to 1) within a custom protocol segment in the network packet header as a truncation signature, or it uses a hash algorithm to generate a message digest for the truncation event as the truncation signature. The first edge node merges the already encapsulated incremental operation vector and the truncation signature into an outgoing buffer queue, awaiting transmission to the second edge node.

[0065] S307, the first edge node packages and merges the incremental operation vector and the truncated feature code into a network data frame through a bidirectional data communication link. The first edge node delivers the network data frame into the sending queue and sends the network data frame to the second edge node deployed in the second region.

[0066] S308, the second edge node continuously listens to and receives network data frames on the local network listening port. The second edge node performs unpacking and parsing of the network data frames to extract truncation signatures. The second edge node determines that the received data is not a full state data packet based on the truncation signature. The second edge node uses the truncation signature as a trigger signal to start the local closed-loop replay calculation process.

[0067] In addition, when the second edge node fails to extract the truncated feature code, i.e., determines that the received data is a full state data message, the second edge node skips the local closed-loop replay calculation process of calling the fixed-point arithmetic engine. Instead, the second edge node directly deserializes the full state data message to extract the state of the first influence subgraph and stores it in the local temporary cache area to wait for subsequent intersection detection calculation.

[0068] S309. For the data frame packet delivery and port listening and receiving operation process in a cross-regional network environment, those skilled in the art can call the underlying socket communication interface of the transmission control protocol to perform data transmission and reception processing. The network socket communication mechanism is a well-known technology in this field and will not be described in detail here.

[0069] S310, when the trigger signal is a truncated feature code, i.e., when it is determined to execute the local closed-loop replay calculation branch, the second edge node reads the node identifier and the state change value after fixed-point number formatting from the received incremental operation vector. The second edge node searches for the independent vertex corresponding to the node identifier in the globally dependent directed acyclic graph copy synchronized in local memory. The second edge node uses the searched and located independent vertex as the starting point for the simulation derivation.

[0070] S311, the second edge node calls the fixed-point number calculation engine integrated inside the second edge node. The second edge node performs spatial interference range simulation calculation according to the derivation rules of the directed acyclic graph that depends on the global project. The derivation rules specifically refer to the execution logic of passing down the state change value layer by layer from the calculation starting point along the direction of the directed edge and superimposing the calculation of the parameter variables of the subsequent node.

[0071] S312, the second edge node uses a fixed-point arithmetic engine to calculate the state of the first influence subgraph through local closed-loop replay. The fixed-point arithmetic engine uses pure integer encoding to eliminate calculation errors caused by differences in the underlying hardware calculation rules of heterogeneous central processing units. The second edge node stores the state of the first influence subgraph calculated by closed-loop replay in a local temporary cache area for subsequent comparison processing.

[0072] Step S40 includes the following sub-steps: S401, the second edge node reads the state of the first influence subgraph from the local temporary cache interval (this state is generated by local closed-loop replay calculation or by full network reception and deserialization), the second edge node parses the independent vertices covered in the state of the first influence subgraph one by one and extracts the corresponding node identifiers, and the second edge node combines all the extracted node identifiers to generate the first vertex identifier set.

[0073] S402, the second edge node scans the queue of write operation instructions in its local running memory that are waiting to be processed. The second edge node extracts the range parameters affected by the local engineering data modification actions that have not yet been merged and written back to the underlying database. At the same time, the second edge node synchronously extracts the current topology depth attribute value corresponding to the local write operation instruction and the logical clock stamp value generated locally based on the global time synchronization protocol. The second edge node independently encapsulates the affected range parameters and the above values ​​in memory to generate a second uncommitted influence subgraph.

[0074] S403, the second edge node separates all independent vertices affected by the local operation from the second uncommitted influence subgraph, and the second edge node extracts the node identifiers of the corresponding independent vertices and assembles them to generate the second vertex identifier set.

[0075] S404, the second edge node performs an intersection detection operation on the state of the first affected subgraph and the second uncommitted affected subgraph. The intersection detection operation specifically refers to the memory data overlap calculation operation performed on the first vertex identifier set and the second vertex identifier set. The second edge node performs the intersection detection operation by executing the following calculation formula: ; In the formula, This represents the result of the intersection detection operation; Represents the set of identifiers for the first vertex; This represents the set of identifiers for the second vertex.

[0076] S405, the process of calculating and executing the element overlap comparison between two data sets in memory. Those skilled in the art can call the hash table deduplication algorithm to compare the element identifier field. The set element comparison and filtering technology is a well-known technology in the field and will not be described in detail here.

[0077] S406, the second edge node reads the intersection detection operation result and performs an empty set comparison. When the intersection detection operation result is an empty set, the second edge node determines that the modification actions of the engineering data generated in different regions have not affected the common decoration engineering entity data.

[0078] S407, when the intersection detection operation result is an empty set, the second edge node performs a lock-free merge operation and saves it to disk. The lock-free merge operation saves it to disk, which means skipping the mutual exclusion lock acquisition step of the underlying database. The second edge node appends the state of the first influential subgraph to the corresponding storage table of the local underlying database. At the same time, the second edge node generates a merge success confirmation message and asynchronously reports the latest merged state to the coordination server through the wide area network bidirectional data communication link. The coordination server then updates the master database, thereby achieving eventual consistency of the global state.

[0079] S408. For the underlying database file page refresh and disk writing process, those skilled in the art can call the asynchronous file write application interface to synchronize the memory state to the disk storage. Database concurrent write persistence technology is a well-known technology in this field and will not be described in detail here.

[0080] S409, when the intersection detection operation result of the second edge node is not an empty set, it is determined that a spatial parameter interference problem has occurred. The second edge node extracts the state of the first influence subgraph that generates the coincident interference and the second unsubmitted influence subgraph to form conflict concurrent data.

[0081] S410, the second edge node sends conflict concurrent data to the coordination server deployed in the central computer room along the bidirectional data communication link. The coordination server receives the conflict concurrent data on the listening port and starts the arbitration compensation calculation process.

[0082] S411, when the intersection detection operation results in an intersection conflict, the coordination server reads the corresponding clock sequence value and node depth level parameter for each of the conflicting operations. The coordination server uses the clock sequence value and node depth level parameter to calculate the priority value, and the priority value determines the queuing order of different operations overwriting the main database.

[0083] Step S50 includes the following sub-steps: S501, the coordinating server parses and extracts the conflicting concurrent data, including the conflicting operations, reads the topological depth attribute value recorded by the independent vertices of the conflicting operations, and synchronously extracts the logical clock stamp value carried by the conflicting operations.

[0084] S502, the coordination server calculates the priority value for conflicting operations. The coordination server multiplies the topology depth attribute value and the logical clock stamp value using a preset weight ratio and adds them together. The coordination server then uses the following formula to obtain the priority value: ; In the formula, Represents priority value; Represents the topology depth weighting coefficient; Represents the numerical value of the topology depth attribute; Represents the weighting coefficients for time series data; Represents the logical clock stamp value.

[0085] In S503, the coordinating server sorts and compares the priority values ​​of different operations in the running memory. The coordinating server determines the operation corresponding to the higher priority value as the successful operation and executes it to overwrite the central database. At the same time, the coordinating server converts the engineering size parameters and spatial coordinate parameters carried by the successful operation into the master database update instruction and writes it to the persistent storage medium. It also sends a global state synchronization instruction to all edge nodes to force the local copy to be overwritten.

[0086] S504, the coordinating server uniformly judges the operations with lower priority values ​​as failed operations, and the coordinating server schedules the failed operations to be temporarily suspended in the background memory waiting queue to wait for compensation processing.

[0087] S505, regarding the execution process of updating and overwriting instructions at the database level, those skilled in the art can send standard data update statements to change the stored values ​​of corresponding fields in the data table. Relational data table field overwriting and updating technology is a well-known technology in this field and will not be described in detail here.

[0088] S506: The coordination server requests an independent memory address segment from the underlying operating system that is not shared with the main business process. The coordination server designates the independent memory address segment as an isolated memory space. The coordination server initializes the computing environment and builds an independent sandbox within the isolated memory space. The independent sandbox blocks the interference of subsequent complex compensation calculations on the normal read and write response mechanism of the main database at the physical memory level.

[0089] S507. For the process of allocating and building memory isolation and sandbox environment at the operating system level, those skilled in the art can call system-level namespace isolation functions or containerization interfaces to achieve the encapsulation of underlying resources. Memory space sandbox isolation operation technology is a well-known technology in this field and will not be described in detail here.

[0090] In S508, the coordinating server extracts the original modification target parameters for failed operations pending in the background memory wait queue. The coordinating server deeply copies the decoration project entity nodes associated with the failed operations and the directed edge data of the connections from the main database. The coordinating server injects the copied data structure copy into the independent sandbox and attaches the original modification target parameters to the corresponding nodes of the data structure copy. Inside the independent sandbox, the coordinating server generates a shadow subgraph from the data structure copy. The shadow subgraph serves as an independent recalculation data base dedicated to the failed operations.

[0091] S509, the coordination server queries the main database to read the latest project status data that has just been overwritten by the successful operation. The latest project status data records the actual available building space size and collision avoidance coordinate extreme values ​​after the successful operation takes effect.

[0092] S510, the coordination server converts the engineering status data corresponding to the successful operation into a mathematical inequality constraint range, the coordination server sets the mathematical inequality constraint range as a rigid constraint boundary condition, the coordination server passes the rigid constraint boundary condition to the shadow subgraph, and the rigid constraint boundary condition forcibly limits the absolute limit of subsequent adjustments and changes of the node parameters inside the shadow subgraph.

[0093] S511, the coordination server uses rigid constraint boundary conditions to define the allowable range of expected parameter modifications for failed operations. The coordination server executes the following formula to establish the boundary constraint interval: ; In the formula, It represents the set of feasible parameters formed by rigid constraint boundary conditions; The values ​​of building parameters within the shadow subgraph that need to be compensated and adjusted; The lower limit of available space parameters derived from the engineering status data corresponding to the successful operation; The upper limit value of the available space parameters is derived from the engineering status data corresponding to the successful operation.

[0094] S512, the coordination server calls the internally integrated topology degradation constraint solving engine. The topology degradation constraint solving engine reads the original modification difference that the failed operation was originally expected to write from inside the shadow subgraph. The topology degradation constraint solving engine uses the original modification difference as the initial iteration benchmark point for parameter exploration.

[0095] S513, the topology degradation constraint solving engine performs step size decay calculations on the initial iteration reference point within the shadow subgraph. The engine finds a safe building parameter range by gradually reducing the parameter changes. This safe range represents a set of spatial coordinates and dimensions that best preserves the design intent of the failed operation while strictly adhering to rigid constraint boundary conditions. The engine calculates the compensation modification amount using the following iteration step size formula: ; In the formula, Representing the The compensation modification difference generated by the secondary downgrade exploration calculation; This represents the difference in the original modifications that the failed operation was expected to write. This represents the preset topology degradation attenuation coefficient, which takes a value between zero and one. This represents the current iteration depth order.

[0096] S514, after each compensation modification difference is obtained, the topology degradation constraint solving engine superimposes the compensation modification difference onto the basic attributes of the target vertices inside the shadow subgraph. The topology degradation constraint solving engine verifies whether the absolute value of the superimposed parameter falls within the feasible range limited by the rigid constraint boundary conditions.

[0097] S515, when the topology degradation constraint solving engine passes the verification, it immediately interrupts the step size decay calculation loop, determines that the parameter solution is successful, and establishes the currently verified compensation modification difference as the final legal compensation parameter.

[0098] S516. For the constraint optimization and iterative approximation calculation process in a multivariable spatial environment, those skilled in the art can call the gradient descent algorithm or the simulated annealing algorithm to search for an approximate optimal solution within the feasible region. The multidimensional parameter constraint heuristic spatial search solution technology is a well-known technology in this field and will not be described in detail here.

[0099] In S517, the coordination server concatenates and reassembles the valid compensation parameters with the corresponding independent vertex identifiers into data packets. At the same time, the coordination server extracts the latest logical clock value generated when the successful operation is written to the main database and uses it as the expected version number. The coordination server encapsulates and generates a compensation operation vector with a unified data format. The coordination server then uses the cross-regional wide area network communication link to distribute the broadcast compensation operation vector to all edge nodes in the system.

[0100] In S518, all edge nodes receive the compensation operation vector at the network communication port. The edge nodes parse the valid compensation parameters and the associated expected version number. Each edge node reads the latest version clock value of the target independent vertex from its local underlying database. The edge nodes then perform concurrent verification and determination calculations using an optimistic locking mechanism, executing the following state control determination formula: ; In the formula, Represents the execution status of the edge node; The latest version clock value represents the local target independent vertex of the edge node; The expected version number carried by the compensation operation vector (i.e., the logical clock value corresponding to the victory operation); Represents the parameters for legitimate compensation; This represents an overwrite operation; The representative refused to cover the issue and resubmitted the operation.

[0101] Edge nodes in When the verification passes, it is determined that the current local state is still consistent with the basic state in the sandbox simulation. The edge node directly modifies the state of the directed acyclic graph of global project dependencies in its local memory and submits it to the local underlying database for overwrite writing, thereby achieving deterministic state synchronization at the system's global level; the edge node in If the system determines that the local parameters have changed, it will refuse to write the current compensation and trigger a re-reporting process for the latest local modified status.

[0102] S519, the topology degradation constraint solving engine running inside the coordination server increments the iteration counter every time it performs step decay calculation. The coordination server reads the current iteration calculation depth order generated by the iteration counter in real time. The current iteration calculation depth order reflects the number of loop execution rounds for parameter exploration inside the shadow subgraph.

[0103] S520: The coordination server reads the preset maximum iteration depth threshold from the underlying configuration file. The maximum iteration depth threshold represents the limit of computing resource overhead and the maximum degree of degradation compromise that engineering parameters can tolerate. The coordination server compares the current iteration computing depth order with the maximum iteration depth threshold.

[0104] S521, when the current iteration calculation depth is greater than or equal to the maximum iteration depth threshold and the verified compensation modification difference has not yet been calculated, the coordination server determines that the current spatial interference state is an irreconcilable absolute conflict on a physical scale. At this time, the coordination server triggers the calculation depth circuit breaker mechanism and executes the following determination formula to output the circuit breaker blocking status flag: ; In the formula, Indicates the circuit breaker's tripping status; numerical value This indicates that a circuit breaker blocking command has been triggered; the value is... This indicates that the computation state can be maintained; This represents the current iteration depth order. This represents the maximum iteration depth threshold preset by the system.

[0105] S522: When the circuit breaker state is marked as triggered, the coordinating server forcibly terminates the subsequent calculation process of the topology degradation constraint solving engine. The coordinating server issues a memory reclamation command to destroy the independent sandbox running environment. The coordinating server returns the previously requested and allocated independent memory address segment to the underlying operating system. The system thereby avoids infinite loop operations in the unsolvable state and excessive consumption of memory resources.

[0106] S523, the coordinating server retrieves the original designer identifier and the source edge node network address that generated the operation from the background memory waiting queue. The coordinating server concatenates the original modification difference that the failed operation is expected to write with the actual disk space available parameters corresponding to the successful operation. The coordinating server generates a manual intervention alarm message containing detailed conflict context information.

[0107] S524. For the process of capturing abnormal events and generating and distributing alarm messages in the server background, those skilled in the art can call the message queue publish-subscribe interface in the event-driven architecture to execute cross-node targeted notifications. The technology of abnormal monitoring and event notification delivery in distributed systems is a well-known technology in this field and will not be described in detail here.

[0108] S525, the coordination server sends manual intervention alarm messages to the source edge nodes along the wide area network communication link. The source edge nodes parse the manual intervention alarm messages and generate a blocking interruption signal at the local application software interface. The blocking interruption signal freezes the subsequent automatic update permissions of the current independent vertex and presents the parameter conflict details comparison interface to the engineering designers. The system forcibly transfers the spatial overlap parameters that cannot be automatically optimized and compensated by the algorithm to the manual offline negotiation correction process.

[0109] Specific application examples: Scene initialization and global data construction (corresponding to S10): The system has a coordination server deployed in location C (central computer room), a first edge node (structural construction team) deployed in location A, and a second edge node (mechanical and electrical pipeline team) deployed in location B.

[0110] The collaborative analysis system reads the global architectural decoration engineering data of the hotel lobby, abstracts the "load-bearing walls (wall thickness parameters)," "ceiling joist size parameters," and "electrical pipeline layout coordinate data" into independent vertices, and generates directed edges according to the spatial support subordinate relationship to construct a global engineering dependent directed acyclic graph.

[0111] Following the spatial arrangement, the load-bearing wall serves as the baseline support vertex, and its topology depth attribute is assigned an initial value of 0; a core electrical conduit node attached to the wall (node ​​identifier: Node) 089 The longest reachable path length is obtained through pathfinding calculation, and its topological depth attribute value is assigned to 3.

[0112] Edge interception and truncation determination (corresponding to S20-S30): The first edge node at location A intercepts the local structural designer's modification of the engineering data, which states that "the load-bearing wall faces outward by 50mm," and extracts the state change value (+50). The first edge node calls the underlying formatting module to convert +50 into a fixed-point number format and combines it with the logical clock stamp value (assuming it is 1500) to encapsulate and generate an incremental operation vector.

[0113] After traversing downwards, the first edge node is stripped to generate the first influence subgraph. The scan statistics show the total number of independent vertices contained in the first influence subgraph. The total number of dependent edges is 15. It is 24.

[0114] Substitute into the formula to calculate the topological out-degree density: ; because If the network serialization cost exceeds the preset threshold (set to 1.2), the first edge node truncates the entire state data packet and sends only the incremental operation vector and truncated feature code to the second edge node in location B.

[0115] After receiving the data, the second edge node triggers local closed-loop replay calculation based on the truncated feature code and calls the fixed-point arithmetic engine to deduce the state of the first influence subgraph.

[0116] Intersection Conflict Detection and Arbitration (corresponding to S40-S501): Meanwhile, the electromechanical designers in location B are attempting to incorporate the aforementioned core electrical pipeline nodes. 089 Move 30mm towards the wall (the original desired coordinates were moved from 60mm from the origin to 30mm). (It is -30), and the logic clock stamp is 1450.

[0117] The second edge node generates a second uncommitted influence subgraph, and the two are then subjected to intersection detection. ; Since the intersection detection result is not an empty set, the second edge node determines this as conflicting concurrent data and reports it to the coordination server in location C.

[0118] Priority determination and sandbox simulation (corresponding to S502-S516): The coordinating server extracts parameters from both sides and calculates priority values. Assume a topology depth weighting coefficient. The weighting coefficient for time series data is -10 (the shallower the depth, the higher the basic priority). The value is 1.

[0119] Location A (walls, =0, (for 1500): ; Location B (pipeline, It is 3. (1450) ; because > Operation A is determined to be a winning operation and is written to the main database; operation B is determined to be a losing operation and is placed in an isolated sandbox in the isolated memory space, and a shadow subgraph is generated.

[0120] The coordination server reads the latest project status data after the successful operation overwrites it, and derives the lower limit value of the available pipeline space parameter. The maximum value is 50 (due to the wall being extended by 50mm). The value is 100. This creates rigid constraint boundary conditions. ; The coordination server starts the topology degradation constraint solving engine within an independent sandbox, and sets the topology degradation attenuation coefficient. The initial iteration reference point is 60mm from the original coordinates, with a value of 0.6. The desired modification is the difference. The value is -30. Perform step-size decay calculation: 0 (original intention): Modification amount is -30, absolute coordinates = 60 - 30 = 30. 30 does not belong to [50, 100], interference conflict.

[0121] 1: The absolute coordinates are 60 - 18 = 42. Since 42 does not belong to [50, 100], the loop continues.

[0122] 2: Absolute coordinates = 60 - 10.8 = 49.2. Failed.

[0123] 3: Absolute coordinates = 60 - 6.48 = 53.52. Verification passed (50 ≤ 53.52 ≤ 100).

[0124] The engine immediately interrupts the loop, using -6.48 as a valid compensation parameter. Current iteration depth calculation order. 3 is less than the maximum iteration depth threshold The value is 10, therefore A value of 0 indicates that the circuit breaker mechanism will not be triggered.

[0125] State distribution and optimistic locking verification (corresponding to S517-S518): The coordinating server broadcast includes valid compensation parameters and the expected version number. The compensation operation vector is 1500. The second edge node at location B uses optimistic locking for verification, since no new modifications have occurred locally. (for 1500), satisfying Determine the action state The local underlying database was successfully overwritten. Automated collaborative repair of global data was completed.

[0126] The experimental verification and effect comparison are as follows: To verify the actual effectiveness of the "topology degradation constraint solving engine" in this invention, the R&D team extracted experimental data from the conflict scenario where pipeline parameters were squeezed by walls in the above application examples. The comparison was between the "traditional forced rollback interception mechanism (i.e., directly rejecting auxiliary material modifications when overlap occurs)" and the "degradation solving mechanism of this invention".

[0127] Under the traditional mechanism, the modification requests of designers in location B are 100% rejected due to spatial overlap, the design progress is stalled and manual intervention is required to redraw; in this invention, the system automatically finds the legal coordinates (53.52mm) that are closest to the original intention within microseconds, retaining about 21.6% of the designer's design intention space, realizing automated flexible merging without manual intervention, and improving collaborative efficiency.

[0128] The solved parameters strictly fall within the rigid constraint boundary conditions. Above all, it avoids physical interference and collisions with the building.

[0129] Reference Appendix Figure 3 , attached Figure 3 This demonstrates the expected building parameter values ​​of the failed operation (mechanical and electrical piping) within the shadow subgraph, in order of different iteration calculation depths. The trajectory changes as shown. The solid line in the figure represents the path through the formula. The calculated and superimposed parameter trajectory, with the dashed line representing the rigid constraint boundary conditions applied by the victory operation (wall). The value is 50. It can be seen that as the iteration order increases, the compensation parameter exhibits a decreasing step size. When the value is 3, the system successfully crosses the boundary dotted line and falls into the legal set of feasible parameters, proving the effectiveness and robustness of the system in solving the problem of cross-regional concurrent data interference.

Claims

1. A collaborative analysis system for cross-regional architectural decoration engineering data, characterized in that, include The coordination server is deployed in the central computer room and multiple edge nodes are deployed in different regions. The multiple edge nodes are divided into a first edge node deployed in a first region and a second edge node deployed in a second region. The first edge node is used to intercept local engineering data modification actions. When the first edge node strips out the first influence subgraph and calculates that the topology out-degree density of the first influence subgraph is greater than the network serialization cost threshold, it is used to forcibly truncate the full state data packet, generate an incremental operation vector and a truncation feature code and send them to the second edge node. The second edge node is used to receive the incremental operation vector and the truncated feature code. The second edge node triggers the call of the internally integrated fixed-point arithmetic engine to calculate the state of the first influence subgraph in local closed-loop replay based on the truncated feature code. The second edge node is further configured to perform an intersection detection operation on the state of the first influence subgraph and the locally generated second unsubmitted influence subgraph, and when the intersection detection operation results in an intersection conflict, the second edge node sends conflict concurrent data to the coordination server. The coordination server is used to receive the conflicting concurrent data and start the topology degradation constraint solving engine within an independent sandbox to find the safe building parameter range. The coordination server generates a compensation operation vector based on the safe building parameter range and broadcasts it to multiple edge nodes.

2. The collaborative analysis system for cross-regional architectural decoration engineering data according to claim 1, characterized in that, The collaborative analysis system is used to read global architectural decoration engineering data and separate it into different decoration engineering entity data. The collaborative analysis system abstracts the different decoration engineering entity data into independent vertices. The collaborative analysis system generates directed edges connecting the independent vertices according to spatial support and subordinate relationships. The collaborative analysis system performs topological association and assembly of all the independent vertices and the directed edges connecting the independent vertices to generate a global engineering dependency directed acyclic graph. The collaborative analysis system assigns topological depth attribute values ​​to the independent vertices in the global engineering dependency directed acyclic graph according to the spatial arrangement order.

3. The collaborative analysis system for cross-regional architectural decoration engineering data according to claim 2, characterized in that, The first edge node intercepts the local engineering data modification action and extracts the state change value. The first edge node integrates a low-level formatting module, which converts the state change value into a fixed-point number format. The first edge node reads the node identifier of the independent vertex that has undergone the modification operation. The first edge node extracts a monotonically increasing sequence to generate a logical clock stamp value. The first edge node concatenates and encapsulates the node identifier, the state change value converted to the fixed-point number format, the current topology depth attribute value, and the logical clock stamp value to generate the incremental operation vector.

4. The collaborative analysis system for cross-regional architectural decoration engineering data according to claim 3, characterized in that, The first edge node is used to search for successor nodes layer by layer within the directed acyclic graph of global engineering dependencies, using the independent vertex corresponding to the node identifier as the starting vertex. The first edge node merges and separates the starting vertex and all the found successor nodes to generate the first influence subgraph. The first edge node calculates the topology out-degree density of the first influence subgraph so that when the topology out-degree density is greater than the network serialization cost threshold, the first edge node forcibly truncates the first influence subgraph and packages it into a full state data packet to generate the truncation feature code.

5. The collaborative analysis system for cross-regional architectural decoration engineering data according to claim 4, characterized in that, The second edge node integrates the fixed-point number arithmetic engine. The second edge node is used to retrieve and query the independent vertex corresponding to the node identifier in the copy of the global project dependency directed acyclic graph synchronized in local memory as the starting point for calculation. The second edge node calls the fixed-point number arithmetic engine to calculate the state of the first influence subgraph by local closed-loop replay according to the derivation rules.

6. The collaborative analysis system for cross-regional architectural decoration engineering data according to claim 5, characterized in that, The second edge node is used to extract the range parameters affected by the local engineering data modification action that has not yet been merged and written back to the underlying database, and encapsulate the logical clock stamp value generated locally based on the global time synchronization protocol to generate the second unsubmitted impact subgraph; The second edge node extracts the node identifiers corresponding to the independent vertices covered by the state of the first influence subgraph to generate a first vertex identifier set. The second edge node extracts the node identifiers corresponding to the independent vertices covered by the second unsubmitted influence subgraph to generate a second vertex identifier set. The second edge node performs a memory data overlap calculation operation on the first vertex identifier set and the second vertex identifier set to obtain the intersection detection operation result. When the intersection detection operation results in an intersection conflict, the second edge node extracts the state of the first influence subgraph and the second unsubmitted influence subgraph to form the conflict concurrency data carrying their respective corresponding logical clock stamp values.

7. The collaborative analysis system for cross-regional architectural decoration engineering data according to claim 6, characterized in that, The coordination server is used to parse and extract the conflicting concurrent data containing the conflicting operations. The coordination server multiplies the topology depth attribute value and the corresponding logical clock stamp value of the conflicting operation by a preset topology depth weight allocation coefficient and a time series weight allocation coefficient, and adds them to obtain a priority value. The coordination server determines the operation corresponding to the higher priority value as the winning operation, and uniformly determines the operation with the lower priority value as the losing operation. The coordination server requests and allocates an independent memory address segment as an isolated memory space. The coordination server builds an independent sandbox within the isolated memory space. Within the independent sandbox, the coordination server generates a shadow subgraph by copying the data structure of the decoration project entity node associated with the losing operation from the main database.

8. The collaborative analysis system for cross-regional architectural decoration engineering data according to claim 7, characterized in that, The coordination server is used to convert the engineering state data corresponding to the victory operation into mathematical inequality limits and set them as rigid constraint boundary conditions. The coordination server transmits the rigid constraint boundary conditions to the shadow subgraph. The coordination server starts the topology degradation constraint solving engine inside the independent sandbox. The topology degradation constraint solving engine reads the original modification difference that the failed operation was originally expected to write from inside the shadow subgraph as the initial iteration reference point for parameter exploration, and performs step size decay calculation in combination with the rigid constraint boundary conditions to find the safe building parameter range. The topology degradation constraint solving engine establishes the compensation modification difference generated by the step size decay calculation as the final legal compensation parameter.

9. The collaborative analysis system for cross-regional architectural decoration engineering data according to claim 8, characterized in that, The coordination server is used to concatenate and reassemble the legitimate compensation parameters, the corresponding node identifier, and the latest logical clock value corresponding to the victory operation to generate the compensation operation vector carrying the expected version number. Multiple edge nodes receive the compensation operation vector and parse the legitimate compensation parameters and the associated expected version number. Multiple edge nodes use an optimistic locking mechanism to perform concurrent verification and judgment calculation. When the concurrent verification and judgment calculation passes, multiple edge nodes submit the data to the underlying database to perform overwrite write.

10. The collaborative analysis system for cross-regional architectural decoration engineering data according to claim 8, characterized in that, The topology degradation constraint solving engine increments the iteration counter each time it performs the step size decay calculation. The coordination server reads the current iteration calculation depth order generated by the iteration counter in real time. The coordination server compares the current iteration calculation depth order with the maximum iteration depth threshold. When the current iteration calculation depth order is greater than or equal to the maximum iteration depth threshold and the legal compensation parameter has not yet been established, the coordination server triggers a calculation depth circuit breaker mechanism to forcibly terminate the subsequent calculation process of the topology degradation constraint solving engine. The coordination server generates a manual intervention alarm message containing conflict context information and sends it to the source edge node to force a manual offline negotiation and correction process.

Images