Intelligent planning method and system for smart park low-voltage system

Abstract

The application discloses an intelligent planning method and system for a weak current system of a smart park, and belongs to the technical field of weak current intelligent design. The method comprises the following steps: performing feature extraction on weak current multi-source data of the smart park to form a weak current feature set; constructing a multi-dimensional semantic space and obtaining semantic association features according to the weak current feature set, and determining node features, topological relations and constraint rules of the weak current system; generating a weak current knowledge graph based on the information, performing semantic alignment on park basic information, and obtaining a final scene demand representation; performing graph reasoning and constraint calculation according to the representation, obtaining a feasible region and constraint satisfaction condition, and generating a candidate construction scheme; and filtering out an optimal construction scheme from the candidate scheme by using a preset comprehensive optimization strategy and sending the optimal construction scheme to a control center. The scheme promotes the weak current scheme from demand understanding to scheme optimization, forms a coherent and verifiable automatic process, reduces manual intervention, makes design judgment more accurate, and finally outputs a construction scheme with higher engineering reliability.

Description

Technical Field

[0001] This application belongs to the field of low-voltage intelligent design technology, specifically relating to an intelligent planning method and system for a smart park low-voltage system. Background Technology

[0002] In the wave of smart park development, the low-voltage system, as the cornerstone of intelligence, directly impacts the park's operational efficiency and management effectiveness through its planning level. However, with the surge in IoT devices and the iteration of business scenarios, traditional planning methods struggle to adapt to dynamic needs, easily leading to resource waste and hindering the park's digital transformation. Developing data-driven, intelligence-based low-voltage system planning methods has become an urgent need to enhance the park's intelligence.

[0003] Currently, data collection primarily relies on manual inspections or the deployment of fixed sensors to gather operational data from low-voltage equipment and basic information about the park. In the feature extraction phase, features are extracted from the collected data according to pre-defined rules, and a single-dimensional system model is constructed based on this. In the solution generation phase, low-voltage system construction plans are generated based on the constructed model. These plans are then manually screened and evaluated before being finally deployed and applied to the park's low-voltage system.

[0004] However, existing technologies have poor data fusion capabilities and can only process structured data, resulting in insufficient feature extraction. At the same time, optimization decisions rely too much on manual processes and cannot comprehensively consider various implicit factors, making it difficult for the selected solution to achieve global optimality. Summary of the Invention

[0005] To overcome the above-mentioned shortcomings, this invention is proposed to provide solutions or at least partially solve the technical problems of existing technologies, such as poor data fusion capabilities, the inability to process only structured data, resulting in insufficient feature extraction, and excessive reliance on manual optimization decisions, which fail to comprehensively consider various implicit factors, making it difficult for the selected solution to achieve global optimization.

[0006] In a first aspect, the present invention provides an intelligent planning method for a low-voltage electrical system in a smart park, the method comprising:

[0007] Acquire low-voltage multi-source data and basic information of the smart park, extract features from the low-voltage multi-source data, and obtain a low-voltage feature set;

[0008] Based on the set of weak current features, a multidimensional semantic space of the weak current system is constructed, the semantic association features of the multidimensional semantic space are obtained, and the node features, topological relationships between nodes and constraint rules of the weak current system are determined based on the semantic association features.

[0009] Based on the node features, the topological relationships between nodes, and the constraint rules, a low-voltage knowledge graph is generated. Based on the low-voltage knowledge graph, the basic information of the park is semantically aligned to obtain the final scenario requirement representation of the low-voltage system.

[0010] Based on the final scenario requirements, graph reasoning and constraint calculation are performed to obtain the feasible domain and constraint satisfaction conditions of the low-voltage system. Based on the feasible domain and constraint satisfaction conditions, candidate construction schemes for the low-voltage system are generated; wherein, there are at least two candidate construction schemes.

[0011] Based on a preset comprehensive optimization strategy, the optimal construction scheme is selected from all candidate construction schemes, and the optimal construction scheme is sent to the control center.

[0012] In a second aspect, the present invention provides an intelligent planning system for a smart park low-voltage electrical system, the system comprising:

[0013] The data acquisition module is used to acquire low-voltage multi-source data and basic information of the smart park, and to extract features from the low-voltage multi-source data to obtain a low-voltage feature set.

[0014] The semantic modeling module is used to construct a multidimensional semantic space of the weak current system based on the weak current feature set, obtain the semantic association features of the multidimensional semantic space, and determine the node features, topological relationships between nodes, and constraint rules of the weak current system based on the semantic association features.

[0015] The knowledge graph construction module is used to generate a low-voltage knowledge graph based on the node features, the topological relationship between nodes and the constraint rules. Based on the low-voltage knowledge graph, the basic information of the park is semantically aligned to obtain the final scenario requirement representation of the low-voltage system.

[0016] The scheme generation module is used to perform graph reasoning and constraint calculation based on the final scenario requirement representation to obtain the feasible domain and constraint satisfaction conditions of the low-voltage system, and generate candidate construction schemes for the low-voltage system based on the feasible domain and constraint satisfaction conditions; wherein, there are at least two candidate construction schemes.

[0017] The scheme optimization module is used to select the optimal construction scheme from the candidate construction schemes based on a preset comprehensive optimization strategy, and send the optimal construction scheme to the control center.

[0018] In a third aspect, an electronic device is provided, comprising a processor, a memory, and a program or instructions stored in the memory and executable on the processor, wherein the program or instructions are loaded and run by the processor to perform the steps of the intelligent planning method for a smart park low-voltage system described above.

[0019] In a fourth aspect, a computer-readable storage medium is provided, wherein a plurality of program codes are stored therein, the program codes being adapted to be loaded and run by a processor to perform the steps of the intelligent planning method for a smart park low-voltage system described above.

[0020] The above-described technical solutions of the present invention have at least one or more of the following beneficial effects:

[0021] In implementing the technical solution of this invention, the low-voltage electrical solution progresses from understanding the requirements to selecting the best solution, forming a coherent and verifiable automated process; this reduces human intervention, makes design judgments more accurate, and ultimately results in a construction solution with higher engineering reliability. Attached Figure Description

[0022] The disclosure of this invention will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Furthermore, similar numbers in the drawings are used to denote similar components, wherein:

[0023] Figure 1 This is a flowchart illustrating the first main steps of an intelligent planning method for a low-voltage system in a smart park according to an embodiment of the present invention.

[0024] Figure 2 This is a flowchart illustrating the second main step of an intelligent planning method for a smart park low-voltage system according to an embodiment of the present invention.

[0025] Figure 3 This is a schematic diagram of the main structure of an intelligent planning system for a smart park low-voltage system according to an embodiment of the present invention.

[0026] Figure 4 This is a schematic diagram of the structure of an electronic device according to an embodiment of the present invention. Detailed Implementation

[0027] Some embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0028] In the description of this invention, "module" and "processor" can include hardware, software, or a combination of both. A module can include hardware circuitry, various suitable sensors, communication ports, memory, and may also include software components, such as program code, or a combination of software and hardware. A processor can be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and / or signal processing capabilities. The processor can be implemented in software, in hardware, or a combination of both. Non-transitory computer-readable storage media includes any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" means all possible combinations of A and B, such as only A, only B, or A and B. The terms "at least one A or B" or "at least one of A and B" have a similar meaning to "A and / or B" and can include only A, only B, or A and B. The singular terms "a" or "this" can also include plural forms.

[0029] See appendix Figure 1 , Figure 1 This is a flowchart illustrating the first main steps of an intelligent planning method for a low-voltage electrical system in a smart park according to an embodiment of the present invention. Figure 1 As shown, an intelligent planning method for a smart park low-voltage system in an embodiment of the present invention mainly includes the following steps S101-S105.

[0030] Step S101: Obtain low-voltage multi-source data and basic information of the smart park, extract features from the low-voltage multi-source data, and obtain a low-voltage feature set.

[0031] A smart park is a park area that is in the planning stage, awaiting construction, or in the initial construction stage. It covers various functional areas such as offices, scientific research, production, and living, and is an application scenario with digital management goals and information infrastructure planning.

[0032] Low-voltage multi-source data refers to various types of data related to the low-voltage system of the park, collected during the original planning stage. This includes information such as communication cabling planning, power and information interface distribution, security system design, network topology preset, and environmental sensor layout.

[0033] Basic information about the park refers to the basic structural and environmental information related to the park's planning stage. This includes data such as building floor plans, floor layouts, functional zoning, road and pipeline locations, and reserved space for equipment, providing a reference for the design of low-voltage system cabling and equipment selection.

[0034] The low-voltage feature set is a summary of key features extracted from multi-source low-voltage data. It is used to characterize the core attribute information of the park's low-voltage system during the planning stage, such as node type, preset functions, equipment interfaces, wiring requirements, and capacity indicators, and serves the subsequent construction of multi-dimensional semantic space and knowledge graph.

[0035] When collecting low-voltage multi-source data for smart parks, existing planning drawings, design documents, building information models, and preliminary measurement records can be used to collect various types of raw data and digitize them, converting them into a unified, directly processable data format. Basic park information can be collected from building information models, CAD drawings, and planning documents, and organized into structured datasets.

[0036] After obtaining the raw data, feature extraction was performed on the multi-source low-voltage data. First, information on equipment, interfaces, and cabling nodes for each low-voltage system was acquired. Based on the node type, functional attributes, and location coordinates, vectorization was performed to obtain node feature vectors. Next, the pre-defined cabling connections and functional dependencies between nodes were obtained. Based on the connection status and dependency structure between nodes, topological indicators such as degree, path length, and association strength were calculated to form topological feature data between nodes. Then, attribute information such as capacity, number of interfaces, and redundancy requirements for each cabling node and device was acquired. Normalization and standardization were performed on the attribute values ​​and functional requirements to obtain constraint attribute data. Finally, the node feature vectors, inter-node topological feature data, and constraint attribute data were integrated to form a low-voltage feature set for low-voltage system analysis. This set includes the type, function, location, and constraints of each node, as well as the connection relationships and dependencies between nodes.

[0037] Step S102: Construct a multidimensional semantic space for the low-voltage system based on the set of low-voltage features, obtain semantic association features of the multidimensional semantic space, and determine the node features, topological relationships between nodes, and constraint rules of the low-voltage system based on the semantic association features.

[0038] The low-voltage system is a general term for the non-power electrical systems in the park, including security systems, communication systems, information cabling, monitoring and alarm systems, fire alarm systems, network and data center facilities, etc.

[0039] A multidimensional semantic space is a structured semantic model formed by mapping the features, nodes, and dependencies of a low-voltage system to a high-dimensional space. In this space, different dimensions correspond to node attributes, inter-node relationships, functional dependencies, or constraints. The semantic relationships between features can be quantified through distance or similarity in the space.

[0040] Semantic association features are information obtained by analyzing node features and their dependencies in a multidimensional semantic space, specifically used to describe the semantic relationships between nodes. For example, the strength of functional dependencies between nodes, topological similarity, and constraint coupling can all serve as semantic association features to guide the determination of node features and topological relationships.

[0041] Node characteristics are the attribute information of each node in a low-voltage system, such as equipment, interface, and wiring endpoints. These attributes include node type, functional category, physical location, capacity, number of interfaces, redundancy requirements, etc., which are used to describe the basic status and capabilities of the node.

[0042] The topology between nodes is the connection and dependency structure between nodes in a low-voltage system, which includes wiring paths, logical function dependencies, communication links, and direct or indirect connection methods between nodes.

[0043] Constraints are the limitations that must be followed when designing and planning low-voltage systems. These constraints include physical constraints, regulatory constraints, performance constraints, and budget or resource limitations. Physical constraints include things like cabling path length and equipment installation space; regulatory constraints include things like building and electrical standards; and performance constraints include things like network latency and bandwidth requirements.

[0044] After obtaining the low-voltage feature set of the low-voltage system, the first step is to perform structured encoding on this feature set. Numerical features undergo normalization and interval scaling to allow comparison of data with different dimensions; categorical features are represented using one-hot encoding or nested encoding to generate dense representations; time-series features are extracted for variation patterns through sliding windows, frequency domain transformation, or time embedding; and spatial location features are represented using coordinate normalization or location embedding vectors to generate a unified expression. These different processing results are then integrated to form a standardized feature vector set. Mutual information, correlation testing, sparse regularization selection, and tree-based feature importance analysis are then used to filter feature dimensions, removing noise and redundant information to make the feature representation more compact while preserving the system structure's distinctiveness.

[0045] After obtaining the filtered feature vectors, a multidimensional semantic space for the low-voltage system is constructed based on embedding learning methods. Specifically, an initial feature map is first constructed based on the similarity between feature vectors, such as cosine similarity or Gaussian kernel similarity. Then, high-dimensional semantic vectors of nodes are learned through random walk embedding, neighborhood aggregation embedding, or graph attention mechanisms. Commonly used algorithms include DeepWalk, Node2Vec, GraphSAGE, and GAT. This allows the functional dependencies, structural relationships, and implicit constraints of nodes to be quantitatively expressed in the vector space. Based on the formed semantic vectors, the distance, similarity, attention weights, and neighborhood overlap between nodes are further analyzed to extract semantic association features between nodes. Spectral clustering or community detection methods can be used to identify semantically similar or functionally coupled node groups; potential dependencies can be discovered through high-order neighborhood analysis; and constraint coupling regions between nodes can be determined by semantic density. After quantifying these semantic associations, semantic association features with clear dependency strengths and association patterns can be formed.

[0046] Based on the distribution of attribute dimensions in the multidimensional semantic space, the core attribute vectors of nodes are extracted through feature projection. Then, clustering algorithms, such as K-Means, hierarchical clustering, and density clustering, are used to identify different attribute categories, and the center vector or main component of each category is used as the node feature representation. If a node is found to have missing or incomplete attributes, logical inference or rule completion algorithms are used to complete and standardize the attributes, ultimately generating node features with consistent structure.

[0047] Next, topological relationships between nodes are generated based on semantic association features. Specifically, preliminary connections are first established according to an association strength threshold, using path optimization algorithms such as Dijkstra's algorithm or heuristic-based algorithms. The process involves generating feasible physical and logical links; then, combining functional dependencies, determining higher-order logical paths through dependency propagation algorithms; finally, using minimum spanning tree or connectivity analysis, obtaining the topological relationships between nodes. These relationships must reflect both the physical wiring logic and the functional dependency network of the low-voltage system.

[0048] Finally, constraint rules are mined based on the topological relationships between nodes, semantic coupling information, and feature association patterns. Co-occurrence pattern mining algorithms, such as Apriori and FP-growth, are used to identify stable association patterns in the design of low-voltage systems. Rule induction algorithms, such as FOIL or RIPPER, are used to generate logical constraints at the design and operational levels. Then, constraint propagation analysis is used to identify implicit cross-node constraints in the topology, forming constraint rules covering wiring restrictions, resource restrictions, specification restrictions, and performance limits.

[0049] Based on the above technical solution, optionally, a multi-dimensional semantic space of the low-voltage system is constructed based on the set of low-voltage features, semantic association features of the multi-dimensional semantic space are obtained, and node features, topological relationships between nodes, and constraint rules of the low-voltage system are determined based on the semantic association features, including:

[0050] Obtain the dependency and association information between each feature of the weak current feature set, construct an initial association structure based on the dependency and association information, and vectorize the initial association structure to obtain the weak current feature vector;

[0051] Based on the weak current feature vectors, feature correlation analysis is performed to obtain feature dependency data;

[0052] By mapping the weak current feature vectors and feature dependency data to a multidimensional space, the multidimensional semantic space of the weak current system is obtained.

[0053] Semantic association information in a multidimensional semantic space is obtained, and semantic features are extracted based on the semantic association information to obtain the semantic association features of the weak current system.

[0054] Obtain the feature dimension and relation dimension corresponding to the semantic association feature, and perform node attribute parsing based on the feature dimension to obtain the node feature of the weak current system.

[0055] Based on the aforementioned relational dimension, the topological relationships between nodes in the low-voltage system are identified.

[0056] Obtain the constraints corresponding to the topological relationships between the nodes, perform rule parsing on the constraints, and obtain the constraint rules of the weak current system.

[0057] In this solution, dependency and association information is data that describes the interdependencies or influences between weak current features, such as functional dependencies between devices, interface coupling, wiring constraints, etc., which are used to reveal the logical or physical relationships between various features in the system.

[0058] The initial association structure is a preliminary network structure built based on dependency association information. It treats weak current features as nodes and the dependencies between features as edges, forming a topology of relationships between features of the weak current system.

[0059] The weak current feature vector is the result obtained after numerical and vectorization of the initial association structure. Each feature is encoded with a set of numerical values ​​to encode its attributes, dependencies and interrelationships.

[0060] Feature dependency data is quantified relational information obtained by analyzing weak current feature vectors, which can reflect the strength of dependency, priority or functional coupling between features.

[0061] Multidimensional space is a high-dimensional vector space used to represent the characteristics and relationships of weak current systems. Different dimensions correspond to feature attributes, relationships between features, or functional dependencies, and the semantic associations of features are characterized by their positions and distances in the space.

[0062] Semantic association information is data that reflects the semantic correlation or dependency between weak current features in a multi-dimensional space, such as functional similarity, logical association strength, and topological coupling degree, and is used to extract higher-level system semantic features.

[0063] Semantic association features are high-level indicators or attributes extracted from semantic association information, used to describe the relationships between features of low-voltage systems, such as functional dependency strength, compatibility score, and constraint coupling degree.

[0064] Feature dimension is the dimension in a multidimensional space that corresponds to a single feature attribute. It is used to parse node attributes, such as device type, capacity, number of interfaces, etc., and helps to form node features.

[0065] The relational dimension is a dimension in a multidimensional space used to describe the dependencies or connections between features, such as wiring paths, functionally dependent edges, and communication links, and is used to identify the topological relationships between nodes.

[0066] Constraints are the restrictions and rules that govern the topological relationships between nodes, including physical constraints, performance constraints, specification constraints, and resource constraints.

[0067] After obtaining the set of low-voltage features, it is necessary to identify the dependencies between features. First, attribute parsing is performed on each feature, covering equipment type, interface type, wiring path, functional modules, power requirements, bandwidth requirements, etc. Then, this attribute information is standardized into a unified data structure. Next, rule matching is used to compare the possible dependencies between features one by one. For example, if two devices need to share the same power supply or network interface, a power or network dependency is established; if the output of one device is the input of another device, a functional dependency is established; if the wiring paths have spatial overlap or share conduits, a physical constraint dependency is established.

[0068] Building upon rule-based matching, semantic analysis can be employed to transform the names, types, and functional descriptions of features into vector representations. Similarity calculations such as cosine similarity and Euclidean distance are then used to assess the strength of potential associations, further supplementing the dependencies. Simultaneously, historical construction data and industry standards are referenced to verify or enhance the rationality of dependencies, filtering out unrealistic associations. All identified dependencies and their weights collectively constitute the dependency association information between features.

[0069] Subsequently, the dependency and association information is mapped into a graph structure, with weak current features as nodes and dependencies as weighted edges, to build an initial association structure. The edge weights are then quantified to reflect the strength or priority of the dependencies. Next, the initial association structure is vectorized using graph embedding techniques such as Node2Vec and DeepWalk to encode the nodes and their dependencies into high-dimensional vectors, forming the weak current feature vectors.

[0070] After obtaining the low-voltage feature vectors, potential functional couplings or conflicts are identified by calculating the similarity or dependency coupling between node vectors, such as cosine similarity and Pearson correlation coefficient, generating feature dependency data. Then, the low-voltage feature vectors and feature dependency data are mapped to a multidimensional space. Node information can be encoded using t-SNE, UMAP, or Transformer to obtain the multidimensional semantic space of the low-voltage system, allowing the distance and vector relationships between nodes to quantify the semantic associations between features.

[0071] Next, semantic association information is extracted from the multidimensional semantic space, including the functional coupling strength, logical dependencies, and potential conflicts between nodes. Then, clustering analysis or attention mechanisms are used to extract features from this semantic information, yielding the semantic association features of the low-voltage system. These semantic association features are then decomposed into feature dimensions and relationship dimensions. The feature dimensions are used to parse node attributes, obtaining node features; the relationship dimensions are used to identify the topological relationships between nodes. Finally, the constraints embodied in the topological relationships between nodes, such as wiring path length, number of interfaces, bandwidth, delay, and specification requirements, are parsed to form the constraint rules of the low-voltage system.

[0072] In this scheme, by analyzing the dependencies between weak current features and constructing a multi-dimensional semantic space, the accurate extraction of node attributes, topology, and constraint rules is achieved, providing a reliable foundation for subsequent scheme optimization and automated decision-making, and improving planning accuracy and feasibility.

[0073] Step S103: Based on the node features, the topological relationships between nodes, and the constraint rules, generate a low-voltage knowledge graph. Based on the low-voltage knowledge graph, perform semantic alignment on the basic information of the park to obtain the final scenario requirement representation of the low-voltage system.

[0074] The low-voltage knowledge graph is a semantic data model that uses a graph structure to organize knowledge related to the low-voltage system in a park. This graph treats equipment nodes, functional nodes, wiring nodes, interface nodes, etc., in the low-voltage system as entities, and then constructs various types of edges based on the functional dependencies, topological connections, resource consumption, and specification constraints between entities. At the same time, it integrates node characteristics, topological structure, and design constraints to form a computable and reasonable knowledge network.

[0075] The final scenario requirement representation is a structured expression obtained by semantically aligning the basic information of the park with the low-voltage electrical knowledge graph. It accurately reflects the functional requirements, spatial conditions, resource constraints, and safety specifications of the low-voltage electrical system in the target construction scenario, including requirement information after semantic disambiguation, rule constraint mapping, and functional dependency association. Its purpose is to clarify the functional objectives, performance requirements, wiring conditions, and deployment limitations that the low-voltage electrical system must meet in the target scenario.

[0076] After obtaining node features, inter-node topological relationships, and constraint rules, these three types of information need to be integrated into a unified knowledge representation structure. First, node features are transformed into computable entity attribute representations. This is typically achieved through attribute field standardization, semantic vector encoding, and category labeling to construct a semantic feature vector for each low-voltage node. Next, inter-node topological relationships are converted into entity-relationship-entity triple structures, adapting them to the relational expression forms of RDF and ontology semantic frameworks. Simultaneously, engineering constraints such as equipment compatibility, linkage logic, wiring limitations, and capacity conditions are structured into formal rules, which are then described using logic languages ​​such as OWL or SHACL, ensuring these rules are reasonable and verifiable. After completing the construction of entity attributes, relation triples, and rule descriptions, they are uniformly incorporated into a graph data structure, forming the foundational graph set of the low-voltage knowledge graph. To improve the quality and reasoning capabilities of the knowledge graph, semantic enhancement is also required. This involves generating vectorized embeddings of entities and relationships using graph embedding methods such as TransE and RotatE; expanding potential relationships using rule-based reasoning or graph neural network reasoning; and using consistency checking tools to detect and correct constraint conflicts, semantic conflicts, or engineering logic conflicts. After such enhancement, the low-voltage electrical knowledge graph will possess clear structure, explicit semantics, and consistent rules. Once the low-voltage electrical knowledge graph is constructed, the basic information of the industrial park needs to be mapped to the graph's standard semantic space to achieve semantic alignment. This process takes the basic information of the industrial park as input. First, the raw information undergoes field cleaning, unit normalization, named entity recognition, and professional terminology matching to bring the input information to a comparable and standardized state. Then, text semantic encoding models such as BERT / ERNIE are used to obtain semantic vectors of the industrial park information. These vectors are then compared with the entity embedding vectors in the low-voltage electrical knowledge graph to calculate similarity, accurately mapping the park's equipment requirements, network requirements, security requirements, and other content to the corresponding low-voltage electrical entity categories. Then, using structured graph matching algorithms such as VF2 or GNN-based graph matching networks, the spatial structure of the park is aligned with the topology of the low-voltage system, creating a one-to-one structural representation of regional functions, equipment deployment relationships, and system logic. After mapping, the alignment results are verified using engineering logic based on the constraints in the low-voltage knowledge graph. If discrepancies are found between the requirements and the engineering rules—for example, the number of devices exceeds the limit, bandwidth requirements contradict link capabilities, or the linkage logic is invalid—the semantic category is automatically adjusted or the requirement expression is corrected according to the rule system. Finally, the park semantics, low-voltage equipment knowledge, deployment logic, and engineering constraints are integrated to form a unified high-dimensional requirement expression, generating the final scenario requirement representation of the low-voltage system.

[0077] Based on the above technical solution, optionally, a low-voltage knowledge graph is generated based on the node features, the topological relationships between nodes, and the constraint rules. Based on the low-voltage knowledge graph, semantic alignment is performed on the basic information of the park to obtain the final scenario requirement representation of the low-voltage system, including:

[0078] Based on the node features, the topological relationships between nodes, and the constraint rules, semantic extraction is performed to obtain a unified semantic representation.

[0079] A graph structure initialization operation is performed based on a unified semantic representation to obtain a preliminary graph structure. A consistency check is then performed based on the preliminary graph structure to obtain a weak current knowledge graph.

[0080] The basic information of the park is represented by feature vectorization to obtain the park feature vector. Based on the park feature vector and the weak current knowledge graph, entity similarity is calculated to obtain preliminary semantic alignment results.

[0081] Based on the preliminary semantic alignment results, context association and localization are performed to obtain the final semantic alignment results;

[0082] The key entity links for obtaining the final semantic alignment result are then filtered and mapped to obtain a set of key entity links.

[0083] Based on the key entity link set, rule matching and constraint calculation are performed to obtain a preliminary scenario requirement representation. Constraint verification is then performed based on the preliminary scenario requirement representation to obtain the final scenario requirement representation of the low-voltage system.

[0084] In this scheme, the unified semantic representation is a standardized representation of the node characteristics, topological relationships between nodes, and constraint rules of the low-voltage system after semantic extraction. Its purpose is to eliminate semantic differences between different data sources, node types, or constraint descriptions, enabling this information to be understood and processed uniformly.

[0085] The preliminary graph structure is a preliminary network structure obtained after the graph structure initialization operation is completed based on the unified semantic representation. It can reflect the connection of nodes and their relationships.

[0086] The park feature vector is a numerical vector representation of basic park information, such as building layout, spatial zoning, and functional distribution. It is mainly used for matching nodes and relationships in the low-voltage electrical knowledge graph, as well as calculating the similarity between them.

[0087] The preliminary semantic alignment result is a preliminary matching result obtained after calculating the entity similarity between the park's feature vectors and the weak current knowledge graph. It can reflect the potential correspondence between each entity in the park information and the knowledge graph nodes.

[0088] The final semantic alignment result is an accurate matching result obtained after contextual correlation and optimization based on the initial semantic alignment result. It clearly identifies the final semantic correspondence between park entities and weak current knowledge graph nodes, ensuring the accuracy of information association.

[0089] Key entity links are the entity connection paths closely related to scenario requirements identified in the final semantic alignment result. They are used to indicate the critical paths of functional dependencies and information flow, providing core guidance for requirement mapping.

[0090] The critical entity link set is a collection formed by summarizing all identified critical entity links. It is the basis for scenario requirement mapping and constraint calculation.

[0091] The preliminary scenario requirement representation is a structured representation of scenario requirements based on a set of key entity links, obtained through rule matching and constraint calculation. It can initially reflect the system's functional requirements, dependencies between nodes, and constraints, and is the initial form of requirement representation.

[0092] The final scenario requirement representation is the final version obtained after constraining and validating the initial scenario requirement representation. It describes the functional requirements, dependency links, and constraints of the low-voltage system in a smart park.

[0093] First, the node characteristics in the low-voltage system are processed, and the attributes of each node are analyzed one by one, such as equipment type, number of interfaces, redundancy requirements, etc. Quantifiable feature values ​​are extracted, such as capacity and number of interfaces, which are directly quantified, while category information such as equipment type and function category is converted using one-hot encoding or embedded encoding. At the same time, the topological relationship between nodes is analyzed, the physical connection and functional dependency path between nodes are identified, and the connection method, link length, dependency direction and other information are recorded using adjacency list or adjacency matrix. Then, the constraint rules are formalized, and the engineering constraints, performance constraints and specification constraints are parsed into logical predicates or mathematical constraint expressions, and these constraints are associated with the corresponding nodes or edges between nodes.

[0094] By associating node feature vectors with topological relationships and constraints, a preliminary graph structure is formed. Each node's attribute vector serves as a vertex feature, while the topological relationships and constraints between nodes are input to the graph structure as edges or edge attributes, ensuring that each node and edge is explicitly mapped to its corresponding features and constraints. Next, graph embedding operations are performed on this preliminary graph structure using graph neural network methods such as Node2Vec and GraphSAGE. This maps node attributes, edge attributes, and constraints to a unified high-dimensional semantic space, generating a unified semantic representation that allows for comparison and computation of different node types, topological relationships, and constraints on the same semantic scale.

[0095] Graph structure initialization is performed based on a unified semantic representation, treating each node as a vertex and the dependencies between nodes as edges. The entire graph structure is represented by an adjacency matrix or a sparse graph matrix. Simultaneously, loop detection and constraint conflict detection are employed. Figure 1Consistency scoring and other consistency verification algorithms mark and correct possible circular dependencies, conflicting connections, or nodes and edges that do not meet constraints in the graph, ultimately obtaining a weak current knowledge graph. This ensures that node characteristics, topological relationships between nodes, and constraint rules can be correctly represented in the graph, while maintaining logical and physical consistency.

[0096] Subsequently, the basic information of the park is transformed into a park feature vector. First, data such as spatial layout, functional zoning, building height, room capacity, and power supply location are quantified, with missing values ​​filled using interpolation or default values. Then, continuous features are standardized, and categorical features are encoded using one-hot encoding to generate the park feature vector. Using cosine similarity, Euclidean distance, or embedding spatial distance, the similarity between the park feature vector and the node vectors of the low-voltage electrical knowledge graph is calculated to obtain preliminary semantic alignment results. The matching score between each park entity and the knowledge graph node can be quantitatively expressed. To improve matching accuracy, localization is then performed based on contextual association. Specifically, combining node adjacency relationships, functional links, and spatial relative positions, graph convolutional networks or neighborhood weighting methods are used to adjust the matching scores of the preliminary semantic alignment results, generating the final semantic alignment results. This ensures that each park entity accurately corresponds to the most relevant low-voltage electrical knowledge graph node.

[0097] After obtaining the final semantic alignment result, key entity links are identified, which are the node paths necessary to complete a specific function or information flow. Specifically, path search is performed on all possible functional links, for example, using Dijkstra's algorithm or... The algorithm calculates the functional dependency strength and performance index satisfaction of each path, filters out links with strong dependencies, high priority or key functions, forms a set of key entity links, and records the path, start and end nodes and associated constraints in a structured manner.

[0098] Based on the set of key entity links, rule matching and constraint calculations are performed. Specifically, each link and its node attributes are mapped to preset engineering, performance, and specification constraint formulas, such as wiring path length not exceeding the maximum allowable length, bandwidth not less than the minimum requirement, and node capacity not exceeding the maximum supported capacity. Constraint solving algorithms such as logical reasoning, SAT solvers, and linear programming are used to calculate whether the constraints are met. Nodes or links that do not meet the constraints are marked and reconfigured. The preliminary scenario requirement representation is a structured dataset containing each functional link, corresponding node, and constraint conditions. Finally, constraint verification is performed on the preliminary scenario requirement representation. Specifically, the calculation results of each node and link are compared item by item with hard constraints, performance indicators, and safety specifications. By correcting node allocations or link paths that do not meet the conditions, the final scenario requirement representation of the low-voltage system is obtained.

[0099] This solution can uniformly map various node characteristics, topological relationships between nodes, and design constraints in a low-voltage system into a computable semantic space, thereby quantifying and structuring system information, effectively identifying potential conflicts and inconsistencies, and ensuring the matching of design logic and physical constraints.

[0100] Step S104: Perform graph reasoning and constraint calculation based on the final scenario requirement representation to obtain the feasible domain and constraint satisfaction conditions of the weak current system. Based on the feasible domain and constraint satisfaction conditions, generate candidate construction schemes for the weak current system; wherein, there are at least two candidate construction schemes.

[0101] The feasible domain is the set of solutions obtained by reasoning and calculating the final scenario requirements based on the low-voltage knowledge graph and constraint rule system. All solutions in this set can meet the requirements of engineering logic, specification constraints, performance indicators, and resource conditions, and clearly define the entire feasible configuration range of the low-voltage system in terms of structure, parameter configuration, wiring method, and equipment combination. It is the optional range boundary that the low-voltage system construction plan must follow.

[0102] Constraint satisfaction conditions are a set of conditions formed during the planning of low-voltage systems, taking into account all engineering constraints, regulatory requirements, safety standards, performance indicators, and resource limitations. This set is used to verify whether candidate solutions conform to system logic and engineering feasibility. For example, bandwidth must reach a minimum threshold, wiring paths cannot exceed specified lengths, and devices must meet compatibility requirements; these are all specific contents of constraint satisfaction conditions.

[0103] The candidate construction schemes are multiple construction configuration options generated within the feasible domain based on the constraint satisfaction conditions, each of which can meet the deployment requirements of the low-voltage system. Each scheme includes feasible equipment selection, wiring methods, deployment structure, and system configuration combinations, which can achieve the functional objectives specified in the final scenario requirements representation.

[0104] After obtaining the final scenario requirement representation, the first step is to map this requirement into query and reasoning input for a low-voltage electrical knowledge graph. For each functional requirement, spatial location, and performance indicator in the requirement, entity matching and relationship tracing are performed one by one to filter out a subset of the knowledge graph directly related to the requirement. This subset includes relevant device entities, interface entities, functional dependency edges, and applicable constraint rules. Based on this subset, rule-based reasoning and graph neural network inference are run simultaneously. Rule-based reasoning can employ OWL / SHACL logical rules or a forward reasoning engine. These two reasoning methods derive the reachable path and implicit dependencies of the requirement; for example, function A requires a link from device X to switch Y and then to data center Z to be realized. Simultaneously, semantic embedding methods are used to calculate compatibility scores between entities, identifying weakly related or conflicting relationships.

[0105] The structured information obtained from the low-voltage knowledge graph is transformed into a computable set of constraints. First, through constraint propagation and consistency checks, parameter ranges that satisfy all hard constraints and are internally consistent, along with node and link configurations, are selected. These quantified ranges and sets of optional elements constitute the feasible region of the low-voltage system. For each configuration in the feasible region, logical and numerical judgments are used to verify specification clauses, performance limitations, and resource limitations. The judgment results that meet all constraints are recorded; these judgments that satisfy the conditions are the corresponding constraint satisfaction conditions.

[0106] After obtaining the feasible region and the constraint satisfaction conditions, candidate construction schemes are generated using the parameterized representations within the feasible region. These parameterized representations include the set of optional paths for each link, the list of optional equipment models for each node, and the acceptable performance level for each type of equipment. First, heuristic construction and greedy selection methods are used to quickly generate several preliminary scheme samples that satisfy the hard constraints. Then, these samples are used as seeds to input into a multi-objective optimizer to solve the combinatorial problem. Equipment selection, routing, and redundancy strategies are set as decision variables. The objective function covers cost minimization, performance maximization, and reliability optimization. For small-scale problems, integer linear programming or mixed-integer linear programming is used for exact solutions; for large-scale or multi-criteria objective problems, multi-objective evolutionary algorithms, simulated annealing, or local search are used to search for diverse solution sets within the feasible region, ensuring that multiple candidate construction schemes satisfying the constraints are obtained.

[0107] During the solution generation process, rapid feasibility verification and standard compliance checks will be conducted simultaneously to eliminate solutions that violate soft constraints or have excessively high standard risks. Finally, at least two candidate solutions employing complementary strategies will be selected, and equipment lists, cabling routes, topology diagrams, cost estimates, constraint satisfaction annotations, and compliance reports will be exported in a structured format.

[0108] Based on the above technical solution, optionally, graph reasoning and constraint calculation are performed based on the final scenario requirement representation to obtain the feasible region and constraint satisfaction conditions of the low-voltage system. Based on the feasible region and constraint satisfaction conditions, candidate construction schemes for the low-voltage system are generated, including:

[0109] Based on the final scenario requirement representation, the weak current entities, weak current attributes, and weak current relationships associated with the final scenario requirement representation are located in the weak current knowledge graph. Based on the weak current entities, weak current attributes, and weak current relationships, link reasoning is performed to obtain the preliminary functional relationships of the weak current system.

[0110] Based on the preliminary functional relationships, functional reachability reasoning is performed to obtain the functional reachability result set of the weak current system;

[0111] Obtain the weak current constraints corresponding to the functional reachable result set, extract the feasibility index of the weak current constraints, and summarize and organize them based on the feasibility index and preset constraint rules to obtain the constraint set of the weak current system.

[0112] Based on the set of constraints and the set of functional reachable results, a feasibility assessment and parameter range calculation are performed to obtain the feasible domain and constraint satisfaction conditions of the weak current system.

[0113] Based on the feasible domain and the constraint satisfaction conditions, cabling routes are generated and equipment configuration is analyzed to obtain preliminary candidate solutions; wherein, there are at least three preliminary candidate solutions.

[0114] Based on preset constraints, multiple versions of each preliminary candidate scheme are combined and constraint optimization calculations are performed to obtain candidate construction schemes for the low-voltage system.

[0115] In this solution, low-voltage entities are the core, independently identifiable and referential objects within a low-voltage system, including devices, functional components, and logical nodes. Devices include switches, cameras, and access controllers; functional components include video encoding modules and alarm modules; and logical nodes include subsystems and sections. Low-voltage entities are the foundational nodes in the knowledge graph.

[0116] Low-voltage attributes are data fields used to describe the characteristics of low-voltage entities, such as bandwidth, number of ports, power supply method, installation location, protocol type, and latency requirements. As feature values ​​of entities in a knowledge graph, low-voltage attributes define the entity's capabilities and limitations.

[0117] Low-voltage relationships refer to the semantic connections between low-voltage entities, including functional dependencies, topological connections, business logic relationships, and constraint relationships. For example, a camera's dependence on a switch is a functional dependency, the link connection sequence is a topological connection, video streaming to the platform is a business logic relationship, and power supply dependence on PoE is a constraint relationship. Low-voltage relationships constitute the edge structure of a knowledge graph.

[0118] Preliminary functional relationships are system functional links derived from the reasoning of weak current entities, attributes, and relationships, such as "access control → event reporting → security platform" or "camera → switch → core network → storage platform". Its form is a functional implementation path derived from requirements.

[0119] The functional reachability result set comprises all achievable functional links and their reachability states obtained after performing functional reachability reasoning on the initial functional relationships. Reachability states include reachable, partially reachable, constrained reachable, and unreachable. It includes functional paths, dependent components, and corresponding performance or resource requirements.

[0120] Low-voltage constraints are engineering, performance, resource, and regulatory conditions that limit the design, deployment, and operation of low-voltage systems. These include, for example, upper limits on cabling distance, bandwidth requirements, redundancy levels, power supply modes, budget constraints, and standard and regulatory requirements.

[0121] Feasibility metrics are quantitative indicators used to measure whether a low-voltage constraint can be met. These include maximum supported bandwidth, maximum possible cabling distance, minimum latency limit, redundancy coverage, and cost margin. Feasibility metrics are used to determine whether constraints can be met and to what extent they are met.

[0122] Predefined constraint rules are a set of predefined rules used for constraint judgment, consisting of engineering specifications, design rules, and system logic reasoning rules. Examples include "the length of optical fiber cannot exceed X meters", "core equipment must have redundancy", and "equipment links cannot form loops".

[0123] The constraint set is a structured list of constraints compiled based on feasibility indicators and preset constraint rules, which the system must satisfy. It includes numerical ranges, logical conditions, topological restrictions, and resource conditions, among others.

[0124] The preliminary candidate solutions are prototype solutions derived from the feasible region and constraint satisfaction conditions, which can meet basic design requirements, including cabling routing schemes, equipment selection and combination, and link topology. They should include at least three different configurations, such as cost-priority, performance-priority, and redundancy-priority solutions.

[0125] After obtaining the final scenario requirement representation, the system first scans the requirement text word by word and performs semantic analysis. First, it performs word segmentation and part-of-speech tagging, breaking the sentence into a word sequence. Then, it uses a named entity recognition model adapted to the low-voltage field to extract functional words, equipment words, location words, and performance parameters. Functional words include monitoring and access control linkage; equipment words include camera and card reader; location words include computer room and corridor; and performance parameters include resolution, distance, and bandwidth.

[0126] The identified terms then proceed to the terminology normalization stage. The system calls upon a thesaurus containing standard low-voltage terminology, equipment models, and attribute names. For each identified term, it calculates the string edit distance and word vector similarity, and then uses a semantic matching model to determine which entity in the low-voltage knowledge graph it most closely matches. This converts the natural language description into a standardized entity in the graph, such as converting a high-definition camera into an IP-Camera, while simultaneously recording the corresponding attribute slots, such as resolution, frame rate, and power supply method. Next, for the action or condition descriptions in the requirements, such as automatic triggering, linkage upon entry, or synchronization with lighting, dependency parsing is performed. Through dependency trees, the subject entity, object entity, and triggering condition involved in the action are identified, constructing a structured ternary fragment of entity-attribute-action.

[0127] Based on these standardized entities and action fragments, the system performs neighborhood retrieval in the low-voltage knowledge graph. First, it finds first-order adjacency relationships based on entity IDs and extracts all relevant device-type nodes, attribute-type nodes, and functional relationship edges. Then, it combines the actions, constraints, or conditions in the requirements to filter semantically matched relationships in the neighborhood, such as being triggered, providing video streams, relying on power supply, and linkage control. It also uses graph embedding vectors to calculate the compatibility score between entities and relationships, retaining high-scoring relationships as candidate low-voltage relationships.

[0128] Entities with matching scores exceeding the threshold are added to the weak current entity list, normalized attributes are added to the weak current attribute set, and edges in the graph that are consistent with the demand semantics are added to the weak current relationship set. At the same time, missing upstream and downstream nodes between the three are filled in to form a small subgraph that matches the demand semantics, which serves as the starting point for link reasoning.

[0129] With the low-voltage entities, attributes, and relationships matching the final scenario requirements, we begin link reasoning within the low-voltage knowledge graph, starting from these entities. Specifically, we first call the graph adjacency index based on the node ID of each starting entity to extract all outgoing and incoming edges. We then check whether the semantic label of each edge matches the required action, functional intent, or attribute requirement, such as providing video stream, being triggered, relying on power supply, or linkage switch. Next, we use graph embedding vectors to calculate the similarity score between the requirement semantic vector and the candidate edges. The edge with the highest score is taken as the feasible functional extension of the current entity. To ensure the coherence of the reasoning, we continue to recursively search for adjacent relationships for the extended nodes. Using depth-limited graph traversal algorithms, such as DFS with backtracking, we construct reachable functional links starting from the requirement entities, forming a preliminary functional relationship subgraph that includes entity order, relationship category, and triggering logic.

[0130] Next, functional reachability reasoning is performed on the initial functional relationships. Specifically, each functional link in the diagram is mapped to a functional state transition sequence. Logical rules, such as the trigger condition must be met, upstream power supply must be stable, and the signal path cannot be interrupted, are used to perform a feasibility check on each state transition. If a link has an unmet dependency, such as a camera being powered by a switch that does not have PoE capability in the diagram, it is marked as unreachable and removed. All links that pass the check are included in the functional reachability result set. Each link comes with complete parameters including the successful path, participating devices, required communication links, trigger timing, and operational dependencies.

[0131] After obtaining the functionally reachable result set, the corresponding low-voltage constraints are extracted for each reachable link. Specifically, the device template of each device node in the link is queried to read the performance limits, such as bandwidth capacity, power margin, and maximum number of linkages; protocol limitations, such as ONVIF compatibility and Modbus address range; and layout limitations, such as maximum cabling distance and installation altitude limit. Then, connection constraints are extracted from the relationship edges, such as maximum cabling impedance, fiber attenuation, and signal level attenuation model. These original constraints are integrated into a multi-dimensional parameter list.

[0132] Then, feasibility indicators are calculated from the constraints. For example, the remaining bandwidth percentage equals the theoretical bandwidth of the switch port minus the camera's bitstream requirement divided by the theoretical bandwidth; the power supply feasibility indicator equals the total PoE power supply capacity minus the power requirements of all link devices divided by the total PoE power supply capacity; and the cabling feasibility indicator equals the maximum deployable distance minus the actual path distance divided by the maximum deployable distance. Each indicator is normalized to a range of 0 to 1, and then categorized and organized according to preset constraint rules to determine whether it belongs to resource constraints, timing constraints, security policy constraints, or protocol compatibility constraints, etc., ultimately forming a structured set of constraint conditions.

[0133] Next, a feasibility assessment is performed using the set of constraints and the set of functional reachable results. Specifically, all indicators for each reachable link are calculated to ensure they meet preset thresholds, such as all indicators being no less than 0.2. If these thresholds are not met, the link is marked as infeasible and removed. The remaining links are then processed using parameter range calculations. The feasible parameter range for each link is derived through range operations. For example, the feasible cabling distance range is calculated as the distance from the minimum necessary distance to the maximum allowable distance of the device, and then the intersection of this distance with the lower limit of the reserved space and the upper limit of the routing plan is taken. The bandwidth, power, delay, and reliability ranges are all calculated in the same way. Finally, the feasible domain of the low-voltage system and the range of parameters that meet the conditions are obtained.

[0134] With the feasible region established, the process begins by generating cabling routes based on the relational topology and analyzing device configurations. Specifically, this starts with shortest path methods based on weighted graphs, such as Dijkstra's algorithm. The optimal cabling path is calculated for each link on the building topology map. Then, recommended equipment models are selected based on equipment type and parameter range, such as switches with bandwidth that meet the range and power supply equipment that meets the power requirements. Next, at least three preliminary candidate schemes are generated according to different optimization preferences, such as cost priority, reliability priority, and maintenance priority. For example, scheme A uses high-bandwidth equipment and the shortest route, scheme B uses medium-power equipment with backup link redundancy, and scheme C uses low-cost equipment and an acceptable suboptimal route. Each scheme includes cabling path, equipment list, parameter range, link reachability score, etc.

[0135] After generating at least two preliminary candidate schemes, these schemes undergo multi-version combination and constraint optimization. Specifically, each scheme is first decomposed into independently recombinable routing segments, equipment segments, and parameter range segments. These decomposed segments are then organized into a set of composable candidate segments. Constraint optimization algorithms, such as integer linear programming, heuristic search, or multi-objective genetic algorithms, are then used to rearrange, interchange, merge, or replace segments within the candidate segment set, exploring the combination space between different schemes. During the combination operation, the system consistently adheres to preset constraint rules, such as power supply capacity limits, cabling distance limits, bandwidth ranges, and link reliability requirements, to perform real-time feasibility verification on the combined segment structure, eliminating combinations that do not meet the constraints. For combinations that still meet the constraints, the system calculates a comprehensive score based on multiple optimization objectives, including lowest cost, highest redundancy, lowest latency, and lowest cabling path complexity. The score determines the globally optimal or near-optimal combination scheme. This optimization process forms a set of candidate construction schemes for the low-voltage system. Each scheme includes complete equipment configuration, cabling path, key parameter ranges, and functional link structure.

[0136] This solution automatically links requirement understanding, reasoning, and solution generation, enabling the rapid identification of feasible paths and the selection of multiple solutions that meet the constraints, making low-voltage electrical design more efficient, accurate, and reliable.

[0137] Based on the above technical solution, optionally, after sending the optimal construction plan to the control center, the method further includes:

[0138] Acquire real-time operating data of the low-voltage system during the initial operation phase, and calculate key operating indicators based on the real-time operating data to obtain a set of key operating indicators;

[0139] The set of key operating indicators is compared with the preset design target to obtain a deviation list and performance bottleneck information.

[0140] The optimal construction plan, deviation list, and performance bottleneck information are input into the preset optimization model to obtain the operation optimization plan of the weak current system, and the operation optimization plan is sent to the control center.

[0141] In this plan, the initial operation phase refers to the initial operating cycle after the low-voltage system is built and put into use. It typically covers the commissioning period after system launch, the stability verification phase, and the early load adaptation period. During this phase, the system's operating status is not yet fully stable, and real business traffic, device coordination relationships, and load patterns will gradually emerge.

[0142] Real-time operational data refers to the continuously collected dynamic monitoring data during the operation of a low-voltage system, including equipment operating status, link quality, security event logs, energy consumption data, and abnormal alarm records. Equipment operating status includes metrics such as CPU utilization, port load, and power status, while link quality includes metrics such as bandwidth, latency, and packet loss rate.

[0143] The set of key operational metrics is a combination of core performance indicators calculated from real-time operational data, used to evaluate the operational health and service carrying capacity of low-voltage systems. Typical indicators include link utilization, average and peak latency, equipment margin, throughput, power consumption fluctuation, failure rate, and alarm density.

[0144] The preset design goals are the performance and reliability targets determined during the system construction phase, including bandwidth planning values, maximum allowable latency, redundancy level, power steady-state indicators, service response time threshold, availability level, and energy consumption limit, for example, the availability level is set to 99.99%.

[0145] The deviation list contains information on the differences between key operating indicators and preset design goals, such as actual bandwidth utilization exceeding the planned value, latency exceeding the set threshold, and equipment load approaching the upper limit.

[0146] Performance bottleneck information is the system performance limiting factors located through deviation analysis, such as congested links, overloaded devices, low-performance nodes, and frequently failing areas, which can clearly point out the parts of the system that most need optimization.

[0147] The preset optimization model is an algorithmic model used to generate system optimization strategies. Common models include mathematical programming models, reinforcement learning models, rule-driven optimization engines, or multi-objective optimization algorithms. This model combines the optimal construction plan, deviation data, and bottleneck information to calculate resource adjustment strategies, path reconfiguration plans, and equipment optimization strategies, while meeting reliability, cost, and constraint conditions.

[0148] An operational optimization plan is a set of executable improvement strategies generated by an optimization model, aimed at enhancing the performance of low-voltage systems in real-world business scenarios. The plan may include link rerouting, bandwidth allocation adjustments, monitoring parameter optimization, power redundancy adjustments, equipment upgrade recommendations, and policy configuration modifications.

[0149] In the initial operation phase of the low-voltage system, the system first collects real-time operational data through deployed sensors, controllers, log collection ports, and network management devices. This data includes device online rate, link bandwidth utilization, critical node latency and jitter, camera frame rate and bitrate, access control and security alarm events, energy consumption data, and environmental parameters. After collection, the data undergoes timestamp calibration, outlier removal, and format standardization, and is then aggregated using a sliding window to generate a computable sequence. Subsequently, based on the processed data, key operational indicators are calculated, such as network stability indicators, communication load indicators, video quality indicators, security event indicators, and environmental coupling indicators. These indicators are then integrated to form a set of key operational indicators.

[0150] The system compares and analyzes key operational metrics against preset design goals, which include requirements for network capacity, video clarity, security incident handling capabilities, energy consumption limits, and environmental stability. A deviation list is generated separately at this stage. Specifically, for each key metric, the system calculates the difference between the measured value and the design target value, the frequency of exceeding thresholds, and trend changes, noting the degree and duration of deviation to form a clear deviation list. Based on this deviation list, performance bottlenecks are further identified, analyzing which metric deviations have the greatest impact on overall system performance. Through causal analysis, dependency tracing, and bottleneck contribution calculation, specific bottlenecks such as insufficient capacity, processing latency, abnormal energy consumption, or substandard video quality in key nodes, links, or devices are pinpointed.

[0151] After the deviation list and performance bottleneck information are clearly defined, the system inputs the optimal construction plan, deviation list, and performance bottleneck information into a preset optimization model, and generates an operational optimization plan through the model. The operational optimization plan is then sent to the control center in the form of an instruction set.

[0152] The training process for the pre-defined optimization model is as follows:

[0153] First, a dataset is constructed using historical data. Each historical record contains three types of input information: the optimal construction plan, a list of deviations, and performance bottleneck information. The corresponding labels are the actual implemented optimization plans. During the training phase, the input information undergoes structuring and feature processing. Specifically, the equipment configuration, wiring routing, and parameter information in the optimal construction plan are decomposed into quantifiable features; the differences in various indicators in the deviation list are organized into standardized numerical or categorical features; and the bottleneck location, impact level, and other content in the performance bottleneck information are transformed into feature forms that the model can recognize, ensuring that this information can be effectively understood and processed by the model. Then, supervised learning methods are used for training, such as deep neural networks, gradient boosting trees, or multi-objective regression models, feeding the model with the processed historical data. The core of training is to minimize the error between the model's predicted output and the actual optimized plan. By iteratively adjusting the model parameters, the model learns to accurately map the optimal construction plan, deviation list, and performance bottleneck information to the appropriate optimized plan. During training, techniques such as cross-validation, early stopping, and regularization are used to prevent overfitting. Cross-validation verifies the model's generalization ability by dividing the data into multiple sets. Early stopping terminates training when the model's performance no longer improves. Regularization constrains the model's parameter size to avoid overfitting the training data. After training, the model is validated using independent historical data to ensure that it can stably generate accurate optimization schemes when faced with new optimal construction schemes, bias lists, and performance bottleneck information.

[0154] This solution can reflect the system status in real time, identify performance bottlenecks, and generate optimization strategies based on the optimal construction plan, deviation list, and bottleneck information. It enables coordinated adjustment of equipment and parameters, improves system stability and resource utilization, and reduces the risk of failure.

[0155] Step S105: Based on the preset comprehensive optimization strategy, select the optimal construction scheme from the candidate construction schemes and send the optimal construction scheme to the control center.

[0156] The pre-defined comprehensive optimization strategy is a set of optimization principles and algorithmic rules determined in advance during the selection of low-voltage system construction schemes. Its core is to comprehensively consider various design objectives and constraints, such as cost, performance, reliability, redundancy, and energy consumption. It may include weight allocation, priority ranking, multi-objective optimization methods, and heuristic search rules.

[0157] The optimal construction scheme is the one that performs best in terms of cost, performance, reliability, and other comprehensive indicators, determined from multiple candidate construction schemes after evaluation based on a pre-set comprehensive optimization strategy.

[0158] The control center is the core management platform or entity in a smart park responsible for the centralized management, scheduling, and supervision of the construction and operation of the low-voltage electrical system.

[0159] After obtaining multiple candidate construction schemes, the first step is to map the structured information of each scheme into a computable evaluation vector. This information includes equipment lists, cabling routes, topology relationships, cost estimates, performance indicators, and constraint satisfaction status. Then, different indicators are standardized to provide a basis for comparison among them.

[0160] Next, based on the preset comprehensive optimization strategy, corresponding weights or priorities are assigned to various indicators such as cost, performance, reliability, redundancy, and energy consumption. The comprehensive score of each solution is calculated through a multi-objective scoring function or a weighted sum model. Alternatively, the Pareto optimal ranking method can be used to identify the set of non-dominated solutions, avoiding the omission of potential high-quality solutions when multiple objectives conflict.

[0161] Heuristic search or optimization algorithms are used to further rank and filter candidate solutions. By combining ranking algorithms with constraint filtering, solutions that do not meet hard constraints or specification requirements are eliminated. Then, the feasibility of the remaining solutions under actual operating conditions is verified through simulation or lightweight performance evaluation.

[0162] After obtaining the comprehensive evaluation results, the candidate scheme with the highest comprehensive score or the one that best fits the preset strategy is selected as the optimal construction scheme, and this optimal construction scheme is sent to the control center of the smart park in the form of structured data format or digital report.

[0163] Based on the above steps S101-S105, the low-voltage electrical solution progresses from understanding the requirements to selecting the best solution, forming a coherent and verifiable automated process; this reduces human intervention, makes design judgments more accurate, and ultimately results in a construction solution with higher engineering reliability.

[0164] See appendix Figure 2 , Figure 2 This is a flowchart illustrating the second main step of an intelligent planning method for a low-voltage electrical system in a smart park according to an embodiment of the present invention. Figure 2 As shown, an intelligent planning method for a smart park low-voltage system in an embodiment of the present invention mainly includes the following steps S201-S207.

[0165] Step S201: Obtain low-voltage multi-source data and basic information of the smart park, extract features from the low-voltage multi-source data, and obtain a low-voltage feature set.

[0166] Step S202: Construct a multidimensional semantic space for the low-voltage system based on the set of low-voltage features, obtain semantic association features of the multidimensional semantic space, and determine the node features, topological relationships between nodes, and constraint rules of the low-voltage system based on the semantic association features.

[0167] Step S203: Based on the node features, the topological relationships between nodes, and the constraint rules, generate a low-voltage knowledge graph. Based on the low-voltage knowledge graph, perform semantic alignment on the basic information of the park to obtain the final scenario requirement representation of the low-voltage system.

[0168] Step S204: Perform graph reasoning and constraint calculation based on the final scenario requirement representation to obtain the feasible domain and constraint satisfaction conditions of the weak current system. Based on the feasible domain and constraint satisfaction conditions, generate candidate construction schemes for the weak current system; wherein, there are at least two candidate construction schemes.

[0169] Step S205: Based on the preset comprehensive optimization strategy, select the optimal construction scheme from the candidate construction schemes, perform multi-dimensional simulation analysis on the optimal construction scheme, and obtain the verification results of the construction scheme.

[0170] The verification results of the construction scheme are a quantitative evaluation of the performance, reliability and constraint satisfaction of the optimal construction scheme under various operating scenarios, specifically involving core indicators such as network throughput, latency, redundancy availability, power consumption and security event response.

[0171] First, the equipment configuration, cabling routes, redundancy strategies, and parameter settings from the optimal construction plan are imported into a multi-dimensional simulation environment. Then, scenarios such as different load pressures, equipment failures, and environmental changes are simulated. Using technologies such as discrete event simulation, network traffic simulation, energy consumption model calculation, and security event triggering simulation, multiple test scenarios are run, and the fluctuations of various performance indicators are recorded in real time. At the same time, potential constraint violations are investigated. Finally, all data is summarized and organized to form a complete verification result of the construction plan.

[0172] Step S206: Based on the verification results of the construction scheme, identify potential conflicts and obtain a conflict list.

[0173] The conflict list is a systematic summary of potential resource, performance, or constraint conflicts in the construction plan. For example, situations such as equipment capacity exceeding limits, link bandwidth conflicts, overlapping cabling paths, conflicting redundancy strategies, or conflicts between security and energy consumption goals will all be included.

[0174] First, compare each indicator in the verification results of the construction plan with the preset constraints and dependency rules one by one. Using rule engines, constraint checking algorithms, or logical reasoning, meticulously identify non-compliant indicator combinations and contradictory relationships to pinpoint the source and scope of the conflict. Record each conflict in detail as an entry, clearly indicating the conflict type, involved nodes or links, and the severity of the conflict, compiling a complete conflict list.

[0175] Step S207: Based on the conflict list, modify the scheme to obtain the modified optimal construction scheme, and send the modified optimal construction scheme to the control center.

[0176] When revising a plan based on a conflict list, the system first analyzes the type, scope, and severity of each conflict, clearly marking the affected equipment, links, or parameters as objects to be adjusted. Then, adjustments are made using constraint optimization algorithms and rule-based reasoning. Commonly used constraint optimization algorithms include integer linear programming, mixed integer programming, and multi-objective genetic algorithms. Through these techniques, resources are reallocated, cabling paths are optimized, redundancy strategies are adjusted, or equipment selection is modified for the objects to be adjusted. The core objective is to completely eliminate conflicts while satisfying the original constraints and performance goals, while simultaneously achieving a comprehensive optimization of cost, reliability, and performance. After the plan revision is completed, the system generates a revised optimal construction plan, including an updated equipment list, cabling paths, parameter configurations, and constraint satisfaction status. Finally, this is sent to the control center in the form of structured instructions or reports.

[0177] In this embodiment, potential problems can be identified and corrected in advance, ensuring that the optimal construction plan is safe, feasible and robust, while reducing construction risks and costs.

[0178] Based on the above technical solution, optionally, the solution can be optimized based on the conflict list to obtain a revised optimal construction solution, including:

[0179] Based on the conflict list, the cabling route is corrected to obtain the corrected cabling route scheme;

[0180] Based on the conflict list and the revised cabling routing scheme, the device configuration is corrected to obtain the revised device configuration.

[0181] Cost and performance evaluations are conducted based on the revised cabling routes and equipment configurations to obtain optimized evaluation results. Based on the optimized evaluation results, the revised cabling routes, and the revised equipment configurations, the optimal construction plan is further revised to obtain the revised optimal construction plan.

[0182] In this solution, the cabling routing scheme is the specific layout plan for cables, optical fibers, or network links between various devices in the low-voltage system, including the path, length, type, redundancy strategy, and routing plan of each connection line.

[0183] Equipment configuration is a combination configuration scheme of various devices in a low-voltage system, involving the selection of equipment models, determination of quantity, planning of deployment locations, interface allocation, and setting of operating parameters, while also covering redundancy configuration and function allocation.

[0184] The optimization evaluation results are based on cabling and equipment configuration schemes, and are formed by calculating and scoring the system cost, performance indicators, reliability, and constraint compliance. Performance indicators include latency, bandwidth, and energy consumption.

[0185] During the revision of low-voltage system construction plans, the conflict list is first analyzed to identify potential problems in the cabling route, such as line overlap, insufficient capacity, signal interference, or insufficient redundancy. These issues are then addressed by the system using path planning algorithms, conflict priority ranking, and cabling route simulation verification, based on the cabling route's topology information, node locations, and line constraints. Commonly used path planning algorithms include constrained shortest path search, or those combining capacity constraints... Algorithms such as Dijkstra's algorithm are used. After adjustment, a corrected cabling routing scheme is generated, clearly recording the start and end points, path nodes, and physical connection order of each line.

[0186] The device configuration is revised by combining the conflict list and the corrected cabling routing scheme. First, nodes on each line are matched against the candidate device list, with key considerations including port quantity, power requirements, redundancy requirements, and device performance metrics. Then, constrained optimization algorithms are used to select the most suitable device model and port allocation scheme; algorithms such as integer linear programming, multi-objective genetic algorithms, and heuristic greedy selection are all viable options. Simultaneously, device compatibility, load balancing, and maintenance accessibility are verified. Once confirmed, the revised device configuration is generated, with the device type, quantity, and interface allocation for each node clearly recorded.

[0187] After obtaining the revised cabling routing plan and equipment configuration, a cost and performance evaluation will be conducted. The cost evaluation covers the material costs corresponding to the line length, equipment procurement costs, and subsequent maintenance expenses; the performance evaluation focuses on key indicators such as network bandwidth, communication latency, redundancy availability, power load, and energy consumption. During the evaluation, mathematical models and simulation tools are used to simulate data flow transmission, load distribution, and redundancy switching scenarios, calculating the performance data for each line and each device, ultimately forming a structured optimization evaluation result. This result will detail the cost percentage, specific performance indicators, and constraint fulfillment for each line and device.

[0188] Based on the optimization evaluation results, the revised cabling routing scheme, and equipment configuration, the original optimal construction scheme is revised. According to the evaluation results, conflicting or substandard parts of the scheme are replaced and adjusted, and the overall topology, equipment list, port allocation method, and redundancy strategy are updated to generate a revised optimal construction scheme. The final scheme will fully record cabling routing details, equipment configuration information, cost details, and performance indicators, ensuring that all constraints are met.

[0189] This solution systematically eliminates potential conflicts in cabling routes and equipment configurations, while optimizing cost and performance, making the final construction plan more reliable, feasible, and compliant with design constraints.

[0190] It should be noted that although the steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effects of the present invention, different steps do not necessarily have to be executed in such an order. They can be executed simultaneously (in parallel) or in other orders, and these variations are all within the scope of protection of the present invention.

[0191] Furthermore, the present invention also provides an intelligent planning system for low-voltage electrical systems in smart parks.

[0192] See appendix Figure 3 , Figure 3 This is a main structural block diagram of an intelligent planning system for a smart park low-voltage electrical system according to an embodiment of the present invention. Figure 3 As shown, it specifically includes:

[0193] The data acquisition module 301 is used to acquire low-voltage multi-source data and basic information of the smart park, and to extract features from the low-voltage multi-source data to obtain a low-voltage feature set.

[0194] The semantic modeling module 302 is used to construct a multidimensional semantic space of the weak current system based on the weak current feature set, obtain the semantic association features of the multidimensional semantic space, and determine the node features, topological relationships between nodes, and constraint rules of the weak current system based on the semantic association features.

[0195] The knowledge graph construction module 303 is used to generate a low-voltage knowledge graph based on the node features, the topological relationship between nodes and the constraint rules, and to perform semantic alignment on the basic information of the park based on the low-voltage knowledge graph to obtain the final scenario requirement representation of the low-voltage system.

[0196] The scheme generation module 304 is used to perform graph reasoning and constraint calculation based on the final scenario requirement representation to obtain the feasible domain and constraint satisfaction conditions of the weak current system, and generate candidate construction schemes for the weak current system based on the feasible domain and constraint satisfaction conditions; wherein, there are at least two candidate construction schemes.

[0197] The scheme optimization module 305 is used to select the optimal construction scheme from the candidate construction schemes based on a preset comprehensive optimization strategy, and send the optimal construction scheme to the control center.

[0198] This application provides an intelligent planning system for a smart park's low-voltage electrical system, which can achieve... Figure 1 The various processes implemented in the method implementation examples will not be described again here to avoid repetition.

[0199] Those skilled in the art will understand that all or part of the processes in the method of the above embodiment of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium can include any entity or device capable of carrying the computer program code, a medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content included in the computer-readable storage medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0200] Furthermore, the present invention also provides an electronic device 400, including a processor 401, a memory 402, and a program or instructions stored in the memory 402 and executable on the processor 401. When the program or instructions are executed by the processor 401, they implement the various processes of the above-described intelligent planning method embodiment for a smart park low-voltage system and achieve the same technical effect. To avoid repetition, they will not be described again here.

[0201] It should be noted that the electronic devices in the embodiments of this application include the mobile electronic devices and non-mobile electronic devices described above.

[0202] Furthermore, the present invention also provides a computer-readable storage medium. In one embodiment of the computer-readable storage medium according to the present invention, the computer-readable storage medium can be configured to store a program for executing an intelligent planning method for a smart park low-voltage system according to the above-described method embodiments. This program can be loaded and run by a processor to implement the above-described intelligent planning method for a smart park low-voltage system. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The computer-readable storage medium can be a storage device comprising various electronic devices. Optionally, in the embodiments of the present invention, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0203] Furthermore, it should be understood that since the various modules are only provided to illustrate the functional units of the device of the present invention, the physical devices corresponding to these modules may be the processor itself, or a part of the processor's software, a part of its hardware, or a combination of software and hardware. Therefore, the number of modules shown in the figures is merely illustrative.

[0204] Those skilled in the art will understand that the various modules in the device can be adaptively split or combined. Such splitting or combining of specific modules will not cause the technical solution to deviate from the principles of the present invention; therefore, the technical solutions after splitting or combining will fall within the protection scope of the present invention.

[0205] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A smart planning method for a low-voltage electrical system in a smart park, characterized in that, The method includes: Acquire low-voltage multi-source data and basic information of the smart park, extract features from the low-voltage multi-source data, and obtain a low-voltage feature set; Based on the weak current feature set, a multidimensional semantic space of the weak current system is constructed, semantic association features of the multidimensional semantic space are obtained, and node features, topological relationships between nodes, and constraint rules of the weak current system are determined based on the semantic association features. This includes obtaining dependency association information between each feature of the weak current feature set, constructing an initial association structure based on the dependency association information, and vectorizing the initial association structure to obtain the weak current feature vector. Based on the weak current feature vectors, feature correlation analysis is performed to obtain feature dependency data; By mapping the weak current feature vectors and feature dependency data to a multidimensional space, the multidimensional semantic space of the weak current system is obtained. Semantic association information in a multidimensional semantic space is obtained, and semantic features are extracted based on the semantic association information to obtain the semantic association features of the weak current system. Obtain the feature dimension and relation dimension corresponding to the semantic association feature, and perform node attribute parsing based on the feature dimension to obtain the node feature of the weak current system. Based on the aforementioned relational dimension, the topological relationships between nodes in the low-voltage system are identified. Obtain the constraint conditions corresponding to the topological relationship between the nodes, perform rule parsing on the constraint conditions, and obtain the constraint rules of the weak current system; Based on the node features, the topological relationships between nodes, and the constraint rules, a low-voltage knowledge graph is generated. Based on the low-voltage knowledge graph, the basic information of the park is semantically aligned to obtain the final scenario requirement representation of the low-voltage system. This includes semantic extraction based on the node features, the topological relationships between nodes, and the constraint rules to obtain a unified semantic representation. A graph structure initialization operation is performed based on a unified semantic representation to obtain a preliminary graph structure. A consistency check is then performed based on the preliminary graph structure to obtain a weak current knowledge graph. The basic information of the park is represented by feature vectorization to obtain the park feature vector. Based on the park feature vector and the weak current knowledge graph, entity similarity is calculated to obtain preliminary semantic alignment results. Based on the preliminary semantic alignment results, context association and localization are performed to obtain the final semantic alignment results; The key entity links for obtaining the final semantic alignment result are then filtered and mapped to obtain a set of key entity links. Based on the key entity link set, rule matching and constraint calculation are performed to obtain a preliminary scenario requirement representation. Based on the preliminary scenario requirement representation, constraint verification is performed to obtain the final scenario requirement representation of the low-voltage system. Based on the final scenario requirements, graph reasoning and constraint calculation are performed to obtain the feasible domain and constraint satisfaction conditions of the low-voltage system. Based on the feasible domain and constraint satisfaction conditions, candidate construction schemes for the low-voltage system are generated; wherein, there are at least two candidate construction schemes. Based on a preset comprehensive optimization strategy, the optimal construction scheme is selected from all candidate construction schemes, and the optimal construction scheme is sent to the control center.

2. The intelligent planning method for a smart park low-voltage system according to claim 1, characterized in that, in, Based on the final scenario requirements, graph reasoning and constraint calculation are performed to obtain the feasible region and constraint satisfaction conditions of the low-voltage system. Based on the feasible region and constraint satisfaction conditions, candidate construction schemes for the low-voltage system are generated, including: Based on the final scenario requirement representation, the weak current entities, weak current attributes, and weak current relationships associated with the final scenario requirement representation are located in the weak current knowledge graph. Based on the weak current entities, weak current attributes, and weak current relationships, link reasoning is performed to obtain the preliminary functional relationships of the weak current system. Based on the preliminary functional relationships, functional reachability reasoning is performed to obtain the functional reachability result set of the weak current system; Obtain the weak current constraints corresponding to the functional reachable result set, extract the feasibility index of the weak current constraints, and summarize and organize them based on the feasibility index and preset constraint rules to obtain the constraint set of the weak current system. Based on the set of constraints and the set of functional reachable results, a feasibility assessment and parameter range calculation are performed to obtain the feasible domain and constraint satisfaction conditions of the weak current system. Based on the feasible domain and the constraint satisfaction conditions, cabling routes are generated and equipment configuration is analyzed to obtain preliminary candidate solutions; wherein, there are at least three preliminary candidate solutions. Based on preset constraints, multiple versions of each preliminary candidate scheme are combined and constraint optimization calculations are performed to obtain candidate construction schemes for the low-voltage system.

3. The intelligent planning method for a smart park low-voltage system according to claim 2, characterized in that, in, After sending the optimal construction plan to the control center, the method further includes: Acquire real-time operating data of the low-voltage system during the initial operation phase, and calculate key operating indicators based on the real-time operating data to obtain a set of key operating indicators; The set of key operating indicators is compared with the preset design target to obtain a deviation list and performance bottleneck information. The optimal construction plan, deviation list, and performance bottleneck information are input into the preset optimization model to obtain the operation optimization plan of the weak current system, and the operation optimization plan is sent to the control center.

4. The intelligent planning method for a smart park low-voltage system according to claim 1, characterized in that, in, Before sending the optimal construction plan to the control center, the method further includes: Multidimensional simulation analysis was performed on the optimal construction scheme to obtain the verification results of the construction scheme. Based on the verification results of the construction scheme, potential conflicts are identified to obtain a conflict list; Based on the conflict list, the scheme is modified to obtain the modified optimal construction scheme; Accordingly, the optimal construction plan is sent to the control center, including: The revised optimal construction plan is sent to the control center.

5. The intelligent planning method for a smart park low-voltage system according to claim 4, characterized in that, in, Based on the conflict list, the scheme is optimized to obtain the revised optimal construction scheme, including: Based on the conflict list, the cabling route is corrected to obtain the corrected cabling route scheme; Based on the conflict list and the revised cabling routing scheme, the device configuration is corrected to obtain the revised device configuration. Cost and performance evaluations are conducted based on the revised cabling routes and equipment configurations to obtain optimized evaluation results. Based on the optimized evaluation results, the revised cabling routes, and the revised equipment configurations, the optimal construction plan is further revised to obtain the revised optimal construction plan.

6. An intelligent planning system for a smart park's low-voltage electrical system, characterized in that, The system includes: The data acquisition module is used to acquire low-voltage multi-source data and basic information of the smart park, and to extract features from the low-voltage multi-source data to obtain a low-voltage feature set. The semantic modeling module is used to construct a multidimensional semantic space of the weak current system based on the weak current feature set, obtain semantic association features of the multidimensional semantic space, and determine the node features, topological relationships between nodes, and constraint rules of the weak current system based on the semantic association features; wherein, it includes obtaining the dependency association information between each feature of the weak current feature set, constructing an initial association structure based on the dependency association information, and vectorizing the initial association structure to obtain the weak current feature vector. Based on the weak current feature vectors, feature correlation analysis is performed to obtain feature dependency data; By mapping the weak current feature vectors and feature dependency data to a multidimensional space, the multidimensional semantic space of the weak current system is obtained. Semantic association information in a multidimensional semantic space is obtained, and semantic features are extracted based on the semantic association information to obtain the semantic association features of the weak current system. Obtain the feature dimension and relation dimension corresponding to the semantic association feature, and perform node attribute parsing based on the feature dimension to obtain the node feature of the weak current system. Based on the aforementioned relational dimension, the topological relationships between nodes in the low-voltage system are identified. Obtain the constraint conditions corresponding to the topological relationship between the nodes, perform rule parsing on the constraint conditions, and obtain the constraint rules of the weak current system; The knowledge graph construction module is used to generate a low-voltage knowledge graph based on the node features, the topological relationships between nodes, and the constraint rules. Based on the low-voltage knowledge graph, the basic information of the park is semantically aligned to obtain the final scenario requirement representation of the low-voltage system. This includes semantic extraction based on the node features, the topological relationships between nodes, and the constraint rules to obtain a unified semantic representation. A graph structure initialization operation is performed based on a unified semantic representation to obtain a preliminary graph structure. A consistency check is then performed based on the preliminary graph structure to obtain a weak current knowledge graph. The basic information of the park is represented by feature vectorization to obtain the park feature vector. Based on the park feature vector and the weak current knowledge graph, entity similarity is calculated to obtain preliminary semantic alignment results. Based on the preliminary semantic alignment results, context association and localization are performed to obtain the final semantic alignment results; The key entity links for obtaining the final semantic alignment result are then filtered and mapped to obtain a set of key entity links. Based on the key entity link set, rule matching and constraint calculation are performed to obtain a preliminary scenario requirement representation. Based on the preliminary scenario requirement representation, constraint verification is performed to obtain the final scenario requirement representation of the low-voltage system. The scheme generation module is used to perform graph reasoning and constraint calculation based on the final scenario requirement representation to obtain the feasible domain and constraint satisfaction conditions of the low-voltage system, and generate candidate construction schemes for the low-voltage system based on the feasible domain and constraint satisfaction conditions; wherein, there are at least two candidate construction schemes. The scheme optimization module is used to select the optimal construction scheme from the candidate construction schemes based on a preset comprehensive optimization strategy, and send the optimal construction scheme to the control center.

7. An electronic device comprising a processor, a memory, and a program or instructions stored in the memory and executable on the processor, characterized in that, The program or instructions are adapted to be loaded and run by the processor to perform an intelligent planning method for a smart park low-voltage system according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by a processor to perform an intelligent planning method for a smart park low-voltage system according to any one of claims 1 to 5.

Images