Big model-based machine room electrical scheme intelligent generation and CAD drawing automatic output method

Abstract

The application relates to the technical field of electrical design, in particular to a method for intelligently generating an electrical scheme of a computer room based on a large model and automatically outputting a CAD drawing, which comprises the following steps: standardizing preprocessing and feature extraction of multi-source computer room design demand data, and completing normalization processing and vectorization representation of input data; based on a pre-trained large model, fine tuning is carried out by fusing an electrical design specification data set to construct an electrical design special large model; semantic analysis is carried out on the preprocessed user demand through retrieval enhancement generation technology to generate structured computer room electrical design parameters. The application realizes full-process automation of computer room electrical design from demand input to CAD drawing output, improves design efficiency, and effectively solves the technical problems of low design efficiency, non-uniform standards and errors in traditional manual design.

Description

Technical Field

[0001] This invention relates to the field of electrical design technology, specifically to a method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on large models. Background Technology

[0002] Electrical design of data centers and communication equipment rooms is a core part of the construction process, directly determining the power supply reliability, operational security, and construction costs. Traditional data center electrical design relies heavily on the manual experience of senior designers. Designers need to manually complete tasks such as selecting electrical components, layout planning, cable routing, and drawing CAD drawings based on user needs, building conditions, and national standards and specifications. The entire process is time-consuming and inefficient; typically, the electrical design of a medium-sized data center requires two designers working for 1 to 2 weeks to complete.

[0003] Meanwhile, human error is prone to occur during manual design, and differences in experience among designers can lead to inconsistent design standards. Some design schemes may not comply with national standards and specifications, resulting in potential construction hazards and safety risks later on. In recent years, although some computer-aided electrical design tools have emerged, most of these tools can only assist in drawing and cannot achieve intelligent generation of schemes. Designers still need to manually input a large number of parameters, which cannot fundamentally solve the problems of low design efficiency and reliance on manual labor.

[0004] With the development of large language model technology, some design assistance tools based on large models have emerged. However, most existing technologies can only achieve semantic understanding of natural language, and cannot deeply integrate with professional electrical design specifications, nor can they complete intelligent layout of electrical components, cable path planning, or directly output CAD drawings that conform to industry standards, thus failing to meet the automation needs of data center electrical design. Therefore, we propose a method for intelligent generation of data center electrical solutions and automatic output of CAD drawings based on large models. Summary of the Invention

[0005] The purpose of this invention is to provide a method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on large models, so as to solve the problems mentioned in the background art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: A method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on large models includes the following steps: S1: Standardize and preprocess the multi-source data center design requirements data and extract features to complete the normalization and vectorization representation of the input data; S2: Based on a pre-trained large model, fine-tuning is performed by integrating an electrical design specification dataset to construct a large model specifically for electrical design. Semantic parsing of pre-processed user requirements is performed through retrieval-enhanced generation technology to generate structured electrical design parameters for the computer room. S3: Based on the structured design parameters, an improved constrained KMeans clustering algorithm is used to complete the intelligent layout of electrical components in the computer room, and an improved A* path planning algorithm with process constraints is used to complete the automatic planning of cable paths. S4: Build a multi-dimensional compliance verification engine to verify the national standard compliance of the generated electrical layout and routing schemes. Based on the verification results, drive the large model to iteratively optimize the schemes until the schemes meet all specification constraints. S5: The optimized electrical scheme is mapped to CAD drawing coordinates through a coordinate transformation matrix, converted into parametric CAD drawing instructions, automatically generated computer room electrical CAD drawings that conform to the GB / T6988 drafting standard, and completed standardized output.

[0007] Preferably, in step S1, the multi-source data center design requirement data undergoes standardized preprocessing, including: Collect multi-source input data, including natural language descriptions of requirements, architectural CAD base map data, and historical design case data; The numerical design parameters are subjected to minimum-maximum normalization. The normalization formula is as follows: :in, These are the original parameter values. This is the minimum value of this type of parameter. This is the maximum value of this type of parameter. These are the normalized parameter values; Textual requirements and historical design cases are converted into 768-dimensional high-dimensional vectors using the BERT encoder and stored in a vector database for subsequent retrieval and enhancement generation.

[0008] Preferably, in step S2, the fine-tuning process of constructing the large-scale electrical design model employs a multi-objective loss function, which is: :in, The standard cross-entropy loss is used to constrain the semantic generation capability of large models; To standardize constraint losses, used to ensure that the output of large models conforms to electrical design specifications; The layout constraint loss is used to ensure that the layout parameters output by the large model conform to physical space constraints. and These are the loss weighting coefficients, with values ​​of 0.3 and 0.4 respectively. The normative constraint loss The vector similarity calculation using electrical code clauses is as follows: :in, Vector representations of the output content of large models This is a vector representation of the corresponding electrical design code clauses.

[0009] Preferably, in step S2, the large model adopts the Transformer architecture, and its self-attention mechanism calculation formula is as follows: :in, For querying the matrix, The key matrix, For value matrices, Let be the dimension of the key matrix; The semantic parsing of user requirements is performed through retrieval enhancement generation technology, including: based on the vector representation of user requirements, retrieving the K most similar historical design cases and specification clauses from the vector database, where K is 5; and using the retrieved content as context input to the fine-tuned large model, driving the large model to identify the design intent in the user requirements and extract structured design parameters such as voltage level, load capacity, site size, and safety constraints.

[0010] Preferably, in step S3, an improved constrained KMeans algorithm is used to complete the intelligent layout of electrical components in the computer room. A constrained clustering objective function is constructed based on the power, size, and heat dissipation requirements of the electrical components. :in, The number of clusters. For the first A cluster, As the cluster center, This represents the total area occupied by the elements within the cluster. This represents the available area of ​​the corresponding region. The penalty coefficient is set to 100. By iteratively optimizing the objective function, the electrical components are automatically clustered and laid out, ensuring that the component layout meets space and heat dissipation constraints.

[0011] Preferably, in step S3, the automatic planning of the cable path is completed by combining the improved A* algorithm with process constraints, and its cost function is: ;in, From the starting point of the path to the current node The actual path cost, To start from the current node Estimating the cost of the Euclidean distance to the end point of the path. This represents the constraint cost of the current node, used to characterize the penalty terms for process constraints such as bending radius, cable spacing, and electromagnetic isolation. The constraint weight coefficient is set to 2.0. The constraint cost The calculation uses the generalized cable path evaluation function, the formula of which is: :in, For cable length penalty items, This is a penalty term for the path tilt angle. For cable spacing penalty items, , , These are the weighting coefficients for the corresponding penalty items, with values ​​of 0.5, 0.2, and 0.3 respectively.

[0012] Preferably, in step S3, electrical parameter calculations are performed, including load calculation, voltage drop calculation, and short-circuit current calculation. The formula for calculating three-phase load current is: :in, Active power Line voltage, Power factor; The formula for calculating the voltage drop of a three-phase line is: :in, Resistance per unit length Reactance per unit length This refers to the line length; The formula for calculating the effective value of the periodic component of the three-phase short-circuit current is: :in, Calculate the voltage at the short-circuit point. This is the total impedance of the short-circuit loop.

[0013] Preferably, in step S4, the verification scoring formula of the multi-dimensional compliance verification engine is: :in, The total number of specification clauses to be verified. For the first The verification results of each standard clause, This is an indicator function that takes a value of 1 when the solution meets the condition, and 0 otherwise. This is the weighting coefficient for this clause, with a value ranging from 0 to 1. The weight of critical security clauses is set to 1. The iterative optimization process includes: when the verification score is lower than the preset threshold of 0.95, the specification clauses that fail verification and the problem descriptions are fed back to the electrical design-specific large model, which drives the large model to adjust the layout and path parameters and regenerate the scheme until the verification score reaches the threshold, thus completing the iterative optimization of the scheme.

[0014] Preferably, in step S5, the CAD coordinate transformation uses a homogeneous transformation matrix to transform the world coordinates. Convert to CAD drawing coordinates The conversion formula is: :in, This is the scaling factor. For rotation angle, , It is a translation vector; The objective function for optimizing the scheme is: :in, The total construction cost, This is the total length of the cable. , , These are weighting coefficients, with values ​​of 0.4, 0.3, and 0.3 respectively.

[0015] Preferably, in step S5, the automatic output of CAD drawings includes: The optimized layout coordinates, component parameters, and cable paths are converted into parametric drawing commands in DXF format, automatically generating electrical primary system diagrams and floor plans. Automatically completes dimensioning, component model labeling, and cable specification labeling on drawings, conforming to GB / T6988 drafting standards; The final output is a standard DWG / DXF format CAD file, which also automatically generates a list of equipment and a list of cable quantities.

[0016] Compared with the prior art, the beneficial effects of the present invention are: This invention achieves intelligent conversion of user natural language requirements into structured design parameters by constructing a dedicated large model that integrates electrical design specifications, solving the problem of relying on human experience for requirement understanding in traditional design. Through constrained intelligent layout and path planning algorithms, it realizes the automatic placement of electrical components and cables, taking into account both spatial constraints and process requirements. Through multi-dimensional compliance verification and iterative optimization mechanisms, it ensures that the generated schemes comply with national standards, eliminating the risk of human error. Through coordinate transformation matrices and parametric CAD command generation, it realizes the automatic output from schemes to standard CAD drawings. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the method steps of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] Please see Figure 1A method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on large models includes the following steps: S1: Standardize and preprocess the multi-source data center design requirements data, extract features, and complete the normalization and vectorization of the input data. The specific process is as follows: First, we collect multi-source input data, including design requirements described by users in natural language, CAD base map data provided by the architectural profession, and design case data accumulated by the company in the past. Different preprocessing strategies are adopted for different types of data. For numerical design parameters, such as room area, load capacity, and voltage level, minimum-maximum normalization is performed to map parameters with different dimensions and value ranges to the [0, 1] interval, eliminating the influence of parameter scale differences on subsequent algorithms. The normalization formula is: :in, These are the original parameter values. This is the minimum value of this type of parameter in historical data. This refers to the maximum value of this type of parameter in historical data. These are the normalized parameter values. For example, the range of values ​​for the computer room area parameter is usually 50 to 5000 square meters. After normalization, all area parameters are converted to values ​​between 0 and 1. For text-based requirement descriptions and historical design cases, a pre-trained BERT encoder is used to convert them into 768-dimensional high-dimensional vector representations. The BERT encoder employs a 12-layer Transformer structure, which can fully capture the semantic information and contextual dependencies in the text. All vectorized historical cases and specification clauses are stored in the FAISS vector database to establish an efficient similarity retrieval index, providing a data foundation for subsequent retrieval enhancement generation. For the architectural CAD base map data, the coordinate information and geometric dimensions of building components such as walls, columns, doors, windows, and reserved openings are extracted to create a rasterized map of the computer room space, providing spatial constraints for subsequent layout planning. Specifically, the computer room plan is divided into 100mm×100mm grid units, and each unit is marked as occupiable, unoccupiable, or semi-occupiable, forming the input basis for the layout algorithm.

[0020] S2: Based on a pre-trained large model, fine-tuning is performed using an electrical design specification dataset to construct a dedicated large model for electrical design. Semantic parsing of preprocessed user requirements is then conducted using retrieval-enhanced generation techniques to generate structured electrical design parameters for the computer room. The specific process is as follows: Based on a pre-trained large model using Llama2-70B, and integrating national standard texts such as "GB50052 Design Code for Power Supply and Distribution Systems," "GB50054 Design Code for Low-Voltage Power Distribution," and "GB50174 Design Code for Data Centers," as well as over 5000 historical data center design cases, a dedicated large model for electrical design was constructed through supervised fine-tuning. The fine-tuning process employed a multi-objective loss function. :in, The standard cross-entropy loss is used to constrain the basic semantic generation capabilities of large models. To standardize the constraint loss, the degree of compliance is measured by calculating the cosine similarity between the output content of the large model and the vector of the normative clauses; The layout constraint loss is used to ensure that the layout parameters output by the large model conform to the basic constraints of the physical space. and The loss weight coefficients are set to 0.3 and 0.4 respectively after grid search optimization.

[0021] The large model adopts the standard Transformer architecture, and its self-attention mechanism is calculated using the following formula: :in, For querying the matrix, The key matrix, For value matrices, Let be the dimension of the key matrix; Through multi-head self-attention mechanism, large models can simultaneously capture semantic dependencies in input text from different representation subspaces, and deeply understand the implicit design intent in user needs.

[0022] Retrieval Enhancement Generation (RAG) technology is employed to improve the accuracy of semantic parsing. Based on the vector representation of user needs, the five most similar historical design cases and relevant specification clauses are retrieved from the vector database. This retrieved content, along with the original requirements, is input into a finely tuned large-scale electrical design model as enhanced context. Guided by the retrieved content, the large-scale model accurately identifies key design elements in the user needs, extracts structured design parameters such as voltage level, total load capacity, equipment room size, redundancy requirements, and safety level, and outputs a standardized parameter file in JSON format, providing data input for subsequent layout planning.

[0023] S3: Based on the structured design parameters, an improved constrained KMeans clustering algorithm is used to complete the intelligent layout of electrical components in the computer room, and an improved A* path planning algorithm with process constraints is used to complete the automatic planning of cable paths. The specific process is as follows: Based on the structured design parameters generated in step S2, intelligent layout of electrical components is performed. An improved constrained KMeans clustering algorithm is adopted, and a spatially constrained clustering objective function is constructed based on the characteristics of electrical components, such as power level, physical size, and heat dissipation requirements. :in, This represents the number of clusters, corresponding to different functional areas within the computer room (such as the power distribution room, UPS room, battery room, etc.). For the first A cluster is a set of elements contained in a cluster; The coordinates of the cluster center; This represents the total floor space occupied by all elements within the cluster. This refers to the available area of ​​this functional zone; To constrain the penalty coefficient, a value of 100 is set to ensure the enforceability of the constraint. By iteratively optimizing the objective function, automatic clustering and layout of electrical components such as distribution cabinets, UPS, battery banks, and air conditioning terminals are achieved, ensuring reasonable space utilization in each area and meeting heat dissipation spacing requirements.

[0024] After component placement is completed, an improved A method with process constraints is adopted. The algorithm performs cable path planning. Traditional A... The algorithm only considers path length, which cannot meet the special technological requirements of electrical construction. Therefore, a technological constraint term is introduced into the cost function: :in, This represents the actual path length from the starting point to the current node. The Euclidean distance from the current node to the destination node is used as a heuristic function. The cost of process constraints; The constraint weight coefficient is set to 2.0.

[0025] The process constraint cost is calculated using a generalized cable path evaluation function: :in, A penalty is imposed for cable length to encourage choosing shorter paths; The path tilt angle penalty is applied when the angle between the path and the horizontal or vertical direction exceeds 15 degrees to ensure that the cable is laid in an orthogonal direction as much as possible. This is a cable spacing penalty item, which applies a penalty when the cable spacing for different voltage levels is less than the specification requirement; , , The values ​​are 0.5, 0.2, and 0.3 respectively. The cable paths generated by this algorithm ensure both short paths and meet practical engineering requirements such as bending radius, cable layering, and electromagnetic isolation.

[0026] At the same time, electrical parameters are calculated, and the formula for calculating the three-phase load current is: :in, Active power Line voltage, Power factor; Select an appropriate cable cross-section based on the calculated current. The formula for calculating the voltage drop of a three-phase line is as follows:

[0027] in, Resistance per unit length Reactance per unit length This refers to the line length; Ensure that the voltage drop at the end is within the allowable range specified in the standard (generally not exceeding 5% of the rated voltage). The formula for calculating the effective value of the periodic component of the three-phase short-circuit current is as follows: :in, Calculate the voltage at the short-circuit point. This is the total impedance of the short-circuit loop; Used to verify the breaking capacity of circuit breakers and the dynamic and thermal stability of busbars.

[0028] S4: Construct a multi-dimensional compliance verification engine to verify the national standard compliance of the generated electrical layout and routing schemes. Based on the verification results, drive the large model to iteratively optimize the schemes until the schemes meet all specification constraints. The specific process is as follows: A multi-dimensional compliance verification engine covering over 300 clauses has been built to comprehensively verify the national standard compliance of generated layout and routing schemes. Verification dimensions include: electrical safety distances, equipment heat dissipation spacing, cable laying requirements, fire separation requirements, and redundancy configuration requirements. The verification scoring formula is as follows: :in, This represents the total number of specification clauses to be verified. For the first The verification results of each clause; This is an indicator function; it takes the value 1 if the condition is met and 0 if the condition is not met. As for the weighting of clauses, key clauses involving personal safety (such as electrical clearance and fire protection requirements) are weighted at 1, while general clauses are weighted at 0.3 to 0.7 according to their importance.

[0029] The verification score threshold is set to 0.95. When the score is lower than the threshold, the system automatically organizes the clause number, problem description, and specific violation of the standard clause into natural language and feeds it back to the electrical design big model. The big model adjusts the layout coordinates, path direction, component selection and other parameters according to the feedback information, regenerates the scheme, and then enters the verification process again. This forms a closed-loop iterative mechanism of "generation-verification-optimization" until the verification score reaches 0.95 or above, ensuring that the final scheme fully complies with the national standard requirements.

[0030] S5: The optimized electrical scheme is mapped to CAD drawing coordinates through a coordinate transformation matrix, converted into parametric CAD drawing instructions, automatically generated computer room electrical CAD drawings conforming to the GB / T6988 drafting standard, and standardized output is completed. The specific process is as follows: The optimized electrical scheme is then converted into standard CAD drawings. First, a homogeneous transformation matrix is ​​used to map world coordinates to CAD drawing coordinates: :in, Set the drawing scaling ratio (usually set to 1:100 or 1:50). This is the rotation angle of the drawing (usually 0). , This represents the translation of the origin on the drawing. This transformation matrix ensures that the coordinates in the CAD drawing accurately correspond to the actual building coordinates.

[0031] The overall objective function for optimizing the scheme is: :in, The total construction cost, This is the total length of the cable. , , These are weighting coefficients, with values ​​of 0.4, 0.3, and 0.3 respectively. By minimizing this objective function, an optimal balance is achieved among construction costs, cable usage, and compliance.

[0032] The system converts the transformed coordinates, component parameters, and path information into parametric drawing commands in DXF format, automatically calling the AutoCAD API to complete the drawing. Automatically generated drawings include: primary electrical system diagrams, low-voltage power distribution layout diagrams, UPS system diagrams, and grounding system diagrams. The system automatically completes dimension annotations, component model annotations, cable specification annotations, and circuit number annotations, with all annotation styles strictly conforming to the GB / T6988 standard for the preparation of electrical technical documents.

[0033] The final output is a standard DWG / DXF format CAD file. Simultaneously, it automatically generates an equipment material list and a cable quantity list based on the drawings. These lists include equipment models, quantities, unit prices, total prices, and cable specifications, lengths, and usage information, which can be directly used for budget preparation and procurement. Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on large models, characterized in that, Includes the following steps: S1: Standardize and preprocess the multi-source data center design requirements data and extract features to complete the normalization and vectorization representation of the input data; S2: Based on a pre-trained large model, fine-tuning is performed by integrating an electrical design specification dataset to construct a large model specifically for electrical design. Semantic parsing of pre-processed user requirements is performed through retrieval-enhanced generation technology to generate structured electrical design parameters for the computer room. S3: Based on the structured design parameters, an improved constrained KMeans clustering algorithm is used to complete the intelligent layout of electrical components in the computer room, and an improved A* path planning algorithm with process constraints is used to complete the automatic planning of cable paths. S4: Build a multi-dimensional compliance verification engine to verify the national standard compliance of the generated electrical layout and routing schemes. Based on the verification results, drive the large model to iteratively optimize the schemes until the schemes meet all specification constraints. S5: The optimized electrical scheme is mapped to CAD drawing coordinates through a coordinate transformation matrix, converted into parametric CAD drawing instructions, automatically generated computer room electrical CAD drawings that conform to the GB / T6988 drafting standard, and completed standardized output.

2. The method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on a large model as described in claim 1, characterized in that, In step S1, the multi-source data center design requirement data undergoes standardized preprocessing, including: collecting multi-source input data, which includes natural language descriptions of requirements, architectural CAD base map data, and historical design case data; and performing minimum-maximum normalization on numerical design parameters, using the following normalization formula: ,in, These are the original parameter values. This is the minimum value of this type of parameter. This is the maximum value of this type of parameter. These are the normalized parameter values; textual requirements and historical design cases are converted into 768-dimensional high-dimensional vectors using the BERT encoder and stored in a vector database for subsequent retrieval and enhancement generation.

3. The method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on a large model as described in claim 1, characterized in that, In step S2, the fine-tuning process of constructing a large-scale electrical design model employs a multi-objective loss function, which is: :in, The standard cross-entropy loss is used to constrain the semantic generation capability of large models; To standardize constraint losses, used to ensure that the output of large models conforms to electrical design specifications; The layout constraint loss is used to ensure that the layout parameters output by the large model conform to physical space constraints. and The loss weighting coefficients are 0.3 and 0.4, respectively; the normative constraint loss... The vector similarity calculation using electrical code clauses is as follows: :in, Vector representations of the output content of large models This is a vector representation of the corresponding electrical design code clauses.

4. The method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on a large model as described in claim 1, characterized in that, In step S2, the large model adopts the Transformer architecture, and its self-attention mechanism is calculated using the following formula: :in, For querying the matrix, The key matrix, For value matrices, The dimension of the key matrix is ​​used; semantic parsing of user requirements is performed through retrieval enhancement generation technology, including: based on the vector representation of user requirements, retrieving the K most similar historical design cases and specification clauses from the vector database, where K is 5; the retrieved content is used as context input to the fine-tuned large model, driving the large model to identify the design intent in the user requirements and extract structured design parameters such as voltage level, load capacity, site size, and safety constraints.

5. The method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on a large model as described in claim 1, characterized in that, In step S3, an improved constrained KMeans algorithm is used to complete the intelligent layout of electrical components in the computer room. A constrained clustering objective function is constructed based on the power, size, and heat dissipation requirements of the electrical components. :in, The number of clusters. For the first A cluster, As the cluster center, This represents the total area occupied by the elements within the cluster. This represents the available area of ​​the corresponding region. The penalty coefficient is set to 100; by iteratively optimizing the objective function, the automatic clustering and layout of electrical components are completed, ensuring that the component layout meets the space and heat dissipation constraints.

6. The method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on a large model as described in claim 1, characterized in that, In step S3, the improved A* algorithm with process constraints is used to automatically plan the cable path, and its cost function is: :in, From the starting point of the path to the current node The actual path cost, To start from the current node Estimating the cost of the Euclidean distance to the end point of the path. This represents the constraint cost of the current node, used to characterize the penalty terms for process constraints such as bending radius, cable spacing, and electromagnetic isolation. The constraint weight coefficient is set to 2.0; the constraint cost... The calculation uses the generalized cable path evaluation function, the formula of which is: :in, For cable length penalty items, This is a penalty term for the path tilt angle. For cable spacing penalty items, , , These are the weighting coefficients for the corresponding penalty items, with values ​​of 0.5, 0.2, and 0.3 respectively.

7. The method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on a large model as described in claim 1, characterized in that, In step S3, electrical parameters are calculated, including load calculation, voltage drop calculation, and short-circuit current calculation: the formula for calculating the three-phase load current is: :in, Active power Line voltage, The power factor is used to calculate the voltage drop of a three-phase line. :in, Resistance per unit length Reactance per unit length The line length is given; the formula for calculating the effective value of the periodic component of the three-phase short-circuit current is: :in, Calculate the voltage at the short-circuit point. This is the total impedance of the short-circuit loop.

8. The method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on a large model as described in claim 1, characterized in that, In step S4, the verification scoring formula of the multi-dimensional compliance verification engine is: :in, The total number of specification clauses to be verified. For the first The verification results of each standard clause, This is an indicator function that takes a value of 1 when the solution meets the condition, and 0 otherwise. The weight coefficient for this clause ranges from 0 to 1, with the weight of key safety clauses set to 1. The iterative optimization process includes: when the verification score is lower than the preset threshold of 0.95, the failed verification clauses and problem descriptions are fed back to the electrical design-specific large model, driving the large model to adjust the layout and path parameters, regenerate the scheme, until the verification score reaches the threshold, and the iterative optimization of the scheme is completed.

9. The method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on a large model as described in claim 1, characterized in that, In step S5, the CAD coordinate transformation uses a homogeneous transformation matrix to convert world coordinates. Convert to CAD drawing coordinates The conversion formula is: :in, This is the scaling factor. For rotation angle, , Let be the translation vector; the objective function for optimizing the scheme is: :in, The total construction cost, This is the total length of the cable. , , These are weighting coefficients, with values ​​of 0.4, 0.3, and 0.3 respectively.

10. The method for intelligent generation of computer room electrical schemes and automatic output of CAD drawings based on a large model as described in claim 1, characterized in that, In step S5, the automatic output of CAD drawings includes: The optimized layout coordinates, component parameters, and cable paths are converted into parametric drawing instructions in DXF format, automatically generating electrical primary system diagrams and floor plans; automatically completing dimension annotations, component model annotations, and cable specification annotations on the drawings, conforming to GB / T6988 drafting standards; finally outputting standard DWG / DXF format CAD files, while automatically generating matching equipment lists and cable quantity lists.

Images