AI automatic design system

Abstract

The invention relates to the field of industrial design automation, and particularly discloses an AI automatic design system, which comprises an AI intelligent decision-making and product type selection module used for generating an optimal product type selection and layout strategy based on project requirements; the automatic structure calculation and compliance verification module is used for receiving the strategy, automatically calling external software to carry out calculation and compliance verification, and outputting structure compliance data and a bill of material first draft; the three-dimensional parametric modeling and drawing output module is used for receiving the data to drive three-dimensional software to automatically model and output a drawing and outputting an accurate bill of materials; and the one-stop scheme package generation module is used for integrating all the achievements to automatically generate a complete design scheme package. According to the method, the implicit design experience is encoded into the AI algorithm, automatic deep collaboration of the AI system and the multi-source heterogeneous engineering software is achieved, automation, standardization and global optimization of the whole design process are achieved, and the design efficiency and quality and the knowledge asset protection capacity are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of industrial design automation technology, and more specifically to an AI-based automatic design system. Background Technology

[0002] The design quality of photovoltaic (PV) support systems (including PV trackers, ground-mounted supports, flexible supports, BIPV, BAPV, etc.) directly affects the safety, reliability, and return on investment of PV power plants. Currently, design work in this field relies heavily on manual labor, resulting in several significant technical bottlenecks.

[0003] First, the quality of the design scheme is unstable, heavily reliant on the engineer's personal experience and the current situation, and lacks a standardized and systematic mechanism for generating the optimal solution.

[0004] Secondly, the design process is fragmented and inefficient. From parameter input, structural calculation, 3D modeling to generating a bill of materials, it requires tedious manual operations and data transfer across multiple independent software programs, which is time-consuming.

[0005] More importantly, the implicit experience accumulated by engineers over a long period of practice, which is difficult to express, cannot be effectively preserved and reused, such as selection preferences and cost-performance balance strategies under specific working conditions, resulting in the loss of core knowledge assets as personnel leave.

[0006] Furthermore, in existing technologies, the heterogeneous data formats and closed interfaces between specialized engineering software and enterprise business systems prevent automated instruction scheduling and data flow, becoming a major obstacle to achieving end-to-end automation. Manual design also struggles to quickly find global optimizations under multiple constraints such as cost, materials, and safety, often employing conservative strategies that result in material waste or suboptimal performance.

[0007] Therefore, there is an urgent need for an intelligent design system that can encapsulate design experience, automatically coordinate multiple toolchains, and achieve full-process automation and global optimization. Summary of the Invention

[0008] The purpose of this invention is to provide an AI-powered automatic design system to solve the fundamental problems in the field of photovoltaic support system design, such as low efficiency, unstable design quality, and difficulty in solidifying and passing on core knowledge due to reliance on human experience and fragmented toolchains.

[0009] To solve the above-mentioned technical problems, the present invention specifically provides the following technical solution:

[0010] An AI-powered automatic design system includes:

[0011] The AI-powered intelligent decision-making and product selection module is used to output product selection and layout strategies based on project requirements information, by calling dedicated optimization algorithms and knowledge bases.

[0012] The automated structural calculation and compliance verification module is used to receive the product selection and layout strategy, automatically call the structural calculation software for analysis and verification, and output structural compliance data and a draft bill of materials.

[0013] The 3D parametric modeling and drawing module is used to receive the structural compliance data and the initial draft of the bill of materials, drive the 3D modeling software to automatically model, and output engineering drawings and an accurate bill of materials.

[0014] The one-stop solution package generation module is used to integrate the product selection and layout strategies, structural compliance data, engineering drawings and accurate bill of materials to automatically generate a complete design solution package.

[0015] As a preferred embodiment of the present invention, the AI ​​intelligent decision-making and product selection module specifically includes:

[0016] The information receiving and parsing unit is used to receive project requirement information, and to perform data cleaning and standardization processing on the project requirement information, converting it into a unified structured data format that the system can process;

[0017] The optimization decision-making unit, connected to the information receiving and parsing unit, is used to call a dedicated optimization algorithm and knowledge base for matching and calculation based on the structured project requirement information.

[0018] The strategy generation unit, connected to the optimization decision unit, is used to analyze the calculation results of the optimization decision unit and integrate them to generate the product selection and layout strategy, which includes specific product models, array layouts, installation tilt angles, and spacing parameters.

[0019] As a preferred embodiment of the present invention, the specific workflow of the optimization decision unit is as follows:

[0020] Receive project requirement information in a unified structured data format output by the information receiving and parsing unit;

[0021] Based on the key features in the project requirements information, retrieve matching historical design schemes and applicable expert experience rules from the knowledge base;

[0022] The dedicated optimization algorithm is invoked, with the retrieved historical design schemes as the initial solution and the applicable expert experience rules as constraints, to construct and solve a multi-objective optimization model oriented towards cost, material usage, and structural safety;

[0023] The solution results of the multi-objective optimization model are output as the decision-making basis for generating the product selection and layout strategy.

[0024] As a preferred embodiment of the present invention, the automated structural calculation and compliance verification module specifically includes:

[0025] The task scheduling and parameter injection unit is used to receive the product selection and layout strategy, and automatically call the external structural calculation software according to the strategy to inject the load parameters and geometric parameters required for calculation into the calculation model.

[0026] The automatic compliance verification unit is connected to the calculation task scheduling and parameter injection unit. It is used to obtain the mechanical analysis results output by the structural calculation software and automatically compare and determine the compliance of the results according to the built-in design specifications.

[0027] The bill of materials draft generation unit is connected to the specification compliance automatic verification unit. It is used to generate structured bill of materials data, including component type, specifications, single set usage and material information, based on the mechanical data of the components that have passed the compliance verification and the corresponding product selection information.

[0028] As a preferred embodiment of the present invention, the specific workflow of the computing task scheduling and parameter injection unit is as follows:

[0029] The system receives product selection and layout strategies from the AI ​​intelligent decision-making and product selection module, analyzes the product selection and layout strategies, and extracts product model, installation tilt angle, array spacing and arrangement parameters as basic parameters for calculation input.

[0030] Based on the project's geographical location information in the basic parameters, the corresponding wind load and snow load values ​​are automatically matched from the pre-set standard load library, or the wind load and snow load values ​​are directly calculated using relevant specifications based on the wind and snow data of the project site.

[0031] Integrate the basic parameters with the matched or calculated load data, and call the preset calculation template to generate a standardized calculation input file that conforms to the input format of the target structure calculation software;

[0032] The application programming interface of the target structure calculation software is invoked, and the standardized calculation input file is submitted to start automated structure calculation.

[0033] As a preferred embodiment of the present invention, the specific workflow of the automatic compliance verification unit is as follows:

[0034] Receive the mechanical analysis results output by the structural calculation software, parse the mechanical analysis results, and extract the stress, strain, and displacement data of key components;

[0035] Based on the type and material properties of the key components, the corresponding mandatory standard limits and enterprise safety redundancy thresholds are automatically retrieved from the built-in specification library.

[0036] The extracted stress, strain, and displacement data are compared one by one with the mandatory standard limits and the enterprise safety redundancy threshold, and the compliance of each component with the specifications is determined based on the comparison results.

[0037] The output includes a compliance verification report containing the compliance status of all components, specific data, and the basis for judgment. Component data that are judged to comply with the specifications in the report are marked as mechanical analysis results that have passed the compliance judgment.

[0038] As a preferred embodiment of the present invention, the three-dimensional parametric modeling and plotting module specifically includes:

[0039] The parametric model driving and updating unit is used to receive the structural compliance data and automatically drive the parametric model in the external 3D modeling software to update and reconstruct it based on the geometric parameters and constraints therein.

[0040] The automatic engineering drawing generation and annotation unit is connected to the parametric model driving and updating unit. It is used to automatically generate the front view, top view and side view from the updated 3D model according to the predefined drawing rules, and automatically annotate key dimensions, tolerances and part numbers. Finally, it inserts a standard drawing frame to generate engineering drawings, assembly drawings and part processing drawings that conform to engineering drawing specifications.

[0041] The material information synchronization and bill of materials refinement unit is connected to the engineering drawing automatic generation and annotation unit. It is used to automatically count and extract the material information of all components from the final three-dimensional model, update and refine the initial draft of the bill of materials, and output the accurate bill of materials.

[0042] As a preferred embodiment of the present invention, the specific workflow of the parameterized model driving and updating unit is as follows:

[0043] Receive structural compliance data output by the automated structural calculation and compliance verification module, parse the structural compliance data, and extract the final cross-sectional dimensions, spatial coordinates, and connection angle parameters between the components for each structural component defined therein;

[0044] Call the application programming interface of the 3D modeling software to create a communication session with the preset parametric 3D model. Based on the final cross-sectional dimensions, spatial coordinates and connection angle parameters obtained from the analysis, modify the values ​​of the driving variables of the corresponding features in the preset parametric 3D model in sequence through the communication session.

[0045] The 3D modeling software is triggered to perform a model reconstruction operation based on the updated driving variable values, generating an updated 3D model that perfectly matches the structural compliance data.

[0046] As a preferred embodiment of the present invention, the specific workflow of the material information synchronization and list refinement unit is as follows:

[0047] Receive the updated 3D model from the automatic engineering drawing generation and annotation unit and the initial draft of the bill of materials from the automated structural calculation and compliance verification module;

[0048] Traverse the assembly structure tree of the updated 3D model, and for each component, extract its geometric properties, predefined material properties, and number of instances;

[0049] The extracted geometric and material properties are automatically compared with the corresponding items in the initial draft of the bill of materials to identify specification differences, quantity deviations, or missing items.

[0050] Based on the comparison results, the specifications and parameters in the initial draft of the bill of materials are automatically corrected, the quantities are updated, and missing component information is supplemented. The corrected and supplemented bill of materials is then output as a structured and accurate bill of materials containing precise material information for all components.

[0051] As a preferred embodiment of the present invention, the one-stop solution package generation module specifically includes:

[0052] The document automatic synthesis and typesetting unit is used to receive and integrate the product selection and layout strategy, structural compliance data, engineering drawings and accurate bill of materials, and automatically generate technical document parts including technical specifications and structural calculation reports according to predefined document templates;

[0053] The formatted output and packaging unit, connected to the document automatic synthesis and typesetting unit, is used to standardize and logically associate the technical document, engineering drawings and precise bill of materials, and package them into a unified, deliverable complete design solution package.

[0054] The distribution and system interface unit is connected to the formatted output and packaging unit, and is used to automatically distribute the complete design package to predefined recipients and synchronize the accurate bill of materials to the enterprise resource planning system.

[0055] Compared with the prior art, the present invention has the following advantages:

[0056] 1. By constructing an AI intelligent decision-making module, implicit expert experience is transformed into computable and optimizable algorithms and knowledge bases, achieving standardized and stable output of design quality. It can also automatically seek the global optimal solution under multi-objective constraints, significantly reducing material costs and building knowledge barriers.

[0057] 2. Through the automated structural calculation and verification module, the AI ​​system is deeply and automatically linked with professional engineering software and enterprise information systems, transforming the fragmented manual process into a highly efficient automated production line, and ensuring that the design scheme meets the specifications, thus completely eliminating the risk of human error.

[0058] 3. The system can automatically integrate all design results, generate a complete solution package that can be delivered directly, and automatically synchronize accurate material data to the production and supply chain system, opening up the data flow from design to manufacturing. This not only greatly improves operational and collaborative efficiency, but also creates a new model for the output of technical capabilities. Attached Figure Description

[0059] To more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings in the following description are merely exemplary, and those skilled in the art can derive other embodiments based on the provided drawings without creative effort.

[0060] Figure 1 This is a framework diagram of the system described in Embodiment 1 of the present invention.

[0061] Figure 2 This is a flowchart of the AI ​​intelligent decision-making and product selection module of the system described in Embodiment 1 of the present invention.

[0062] Figure 3 This is a flowchart of the automated structural calculation and compliance verification module of the system described in Embodiment 1 of the present invention.

[0063] Figure 4 This is a flowchart of the three-dimensional parametric modeling and plotting module of the system described in Embodiment 1 of the present invention.

[0064] Figure 5 This is a flowchart of the one-stop solution package generation module of the system described in Embodiment 1 of the present invention. Detailed Implementation

[0065] 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.

[0066] The concepts involved in this application will first be described with reference to the accompanying drawings. It should be noted that the following descriptions of various concepts are only for the purpose of making the content of this application easier to understand and do not constitute a limitation on the scope of protection of this application; furthermore, the embodiments and features in the embodiments of this application can be combined with each other unless otherwise specified. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0067] Example 1

[0068] like Figure 1-5 As shown, the present invention provides an AI-automated design system, comprising:

[0069] The S1.AI Intelligent Decision-Making and Product Selection module is used to output product selection and layout strategies based on project requirements information, utilizing dedicated optimization algorithms and knowledge bases. Specifically, it includes:

[0070] S11. Information Receiving and Parsing Unit, used to receive project requirement information, and perform data cleaning and standardization on the project requirement information, converting it into a unified structured data format that the system can process; specifically:

[0071] This unit serves as the system's front-end processing module, responsible for connecting to external input channels and receiving project requirement information submitted by clients or designers in the form of natural language descriptions, tables, forms, or BIM models.

[0072] Upon receipt, the unit initiates a data cleaning sub-process: First, redundant descriptions, contradictory parameters, and abnormal values ​​that clearly exceed the scope of engineering common sense are removed from the requirement text using regular expression matching and outlier detection algorithms. Then, standardization processing is performed to convert requirement information from different sources and in different formats into a unified structured data format within the system. For example, "Construction location: Zhangjiakou City, Hebei Province" is uniformly encoded as a mapping between a geographical location label and a seismic fortification intensity parameter, and "Installed capacity approximately 50 MW" is standardized into a floating-point numerical field with an added error tolerance label. Finally, a standardized data structure containing fields such as project type, installed capacity, geographical location, environmental constraints, and cost ceiling is output, providing high-quality data input for subsequent units.

[0073] S12. Optimize the decision-making unit, connecting the information receiving and parsing unit. This unit serves as the core computing hub of the AI ​​intelligent decision-making and product selection module. Internally, it employs a layered and progressive processing architecture to implement complex decision-making processes; specifically:

[0074] S121. In the data input phase, the unit receives structured project requirement information from S11. This data structure typically uses a nested key-value pair format and includes multi-dimensional fields such as project type encoding, installed capacity scalar, geographic coordinate vector, environmental parameter matrix, and cost constraint threshold. The unit first initiates the feature engineering sub-process to perform deep analysis and feature extraction on the input data: using principal component analysis (PCA) to extract key feature vectors from the original requirement fields. These vectors can be mathematically represented as point coordinates in a high-dimensional feature space, with each dimension corresponding to a core attribute of the project requirement, such as capacity size, climate severity, and terrain complexity. Simultaneously, a set of constraints is constructed, transforming boundary conditions such as cost ceilings and schedule requirements into the definition of the feasible region hyperplane for the optimization problem.

[0075] S122. During the knowledge base retrieval phase, the unit employs a hybrid retrieval strategy to achieve precise matching. Specifically:

[0076] In the case knowledge layer, a vector retrieval based on cosine similarity is performed. The feature vector of the current project is compared with the feature vector of historical cases in the knowledge base. The top-K historical design schemes with similarity thresholds are retrieved and decoded into an initial solution candidate set in the decision variable space. This candidate set constitutes the initial optimization population in the genetic algorithm framework.

[0077] Logical retrieval based on graph pattern matching is performed at the rule knowledge layer. Environmental parameters and capacity levels in project requirements are used as retrieval conditions. Applicable expert experience rule nodes are located in the rule graph, and these rules are transformed into a set of mathematical constraint equations. For example, the rule "the corrosion protection level of supports in coastal areas should not be lower than C5-M" is transformed into an enumeration constraint of equipment selection decision variables, and the rule "the support height should be increased by 30% in areas where the snow depth exceeds 0.5 meters" is transformed into a conditional branch constraint of structural height variables.

[0078] In addition, the unit also extracts equipment performance parameters from the parameter knowledge layer as the domain boundary of decision variables, such as component power range, bracket load-bearing capacity, inverter efficiency curve, etc., to form the parameter configuration space of the optimization problem.

[0079] S123. During the optimization model construction phase, the unit calls a dedicated optimization algorithm engine, which integrates multiple solvers to adapt to different problem characteristics.

[0080] For the cost objective function, a full life cycle cost model is used for quantification. It not only includes the explicit expenditure of equipment procurement costs, but also internalizes the discounted calculation of installation and commissioning costs, operation and maintenance costs, and residual value recovery. Among them, the installation and commissioning costs have a non-linear monotonically increasing relationship with the bracket type and array complexity, while the operation and maintenance costs are related to the equipment reliability index and the environmental severity level.

[0081] For the objective function of material usage, an approximate calculation system based on the finite element analysis surrogate model is established. This system maps the complex structural stress analysis into a polynomial approximate function of material usage through a proxy model, achieving a balance between calculation efficiency and accuracy. The material usage includes the total mass of the support system, foundation structure, photovoltaic modules and connection nodes, and takes into account the density differences of different materials and specification standardization constraints.

[0082] For the objective function of structural safety, a reliability index system based on limit state theory is constructed. The core indicators are minimizing the maximum stress ratio of the structure and maximizing the limit margin of nodal displacement under extreme wind load and snow load conditions. The fatigue life safety factor is introduced as a measure of long-term reliability. Among them, extreme wind load is the equivalent static load corresponding to the maximum wind speed once in 50 years, snow load is the maximum snow pressure once in 50 years, and the maximum stress ratio is the ratio of actual stress to material yield strength.

[0083] S124. In the multi-objective optimization solution stage, the unit adopts the improved non-dominated sorting genetic algorithm NSGA-II as the master solver. This algorithm has been customized for this application scenario:

[0084] At the coding level, a hybrid coding strategy is adopted, using integer coding for discrete variables such as product model, real number coding for continuous variables such as installation tilt angle, and graph coding based on topological sorting for combined variables such as layout scheme.

[0085] At the genetic operator level, knowledge-guided crossover and mutation operators are designed. The crossover operation prioritizes gene fragment exchange in the similarity dimension of the historical best solution, while the mutation operation performs boundary-limited perturbation based on expert rule constraints to ensure that the newly generated individuals are always within the engineering feasible domain.

[0086] At the strategy selection level, a constraint dominance mechanism is introduced to impose a penalty function on individuals who violate expert experience rules, thereby reducing their fitness value and guiding the population to converge to both the feasible region and the Pareto front.

[0087] During the algorithm iteration process, the unit monitors population diversity indicators in real time, such as the spatial distribution entropy of the solution set, and convergence indicators, such as the improvement rate of the intergenerational objective function. It dynamically adjusts hyperparameters such as crossover probability, mutation probability, and population size to achieve an adaptive optimization process.

[0088] S125. Finally, the unit outputs the solution results of the multi-objective optimization model. This result is not a single optimal solution, but rather a set of non-dominated solutions constituting the Pareto front. Each solution corresponds to a complete parameter configuration for a product selection and layout strategy, including the specific values ​​of decision variables such as the component model selection vector for each subarray, bracket type encoding, installation tilt angle, number of array rows and columns, and spacing parameter matrix, as well as the corresponding objective function values ​​for cost, material usage, and structural safety. The unit also additionally outputs robustness evaluation indicators for each solution, i.e., the sensitivity analysis results of the objective function value changes under demand parameter perturbations, providing multi-dimensional decision-making basis for the strategy generation unit in S13.

[0089] S13. Strategy Generation Unit, connected to the optimization decision-making unit. This unit is responsible for transforming the numerical results of optimization calculations into executable strategy solutions; specifically:

[0090] After receiving the solution results of the multi-objective optimization model output by S12, start the result parsing subprocess:

[0091] The decision variable vector corresponding to the optimal solution is mapped using engineering semantics. The continuously encoded installation tilt angle values ​​are converted into angle settings that conform to the component installation specifications, such as a fixed tilt angle of 18° or a seasonally adjustable tilt angle of 15°-25°. The discretely encoded equipment model index is mapped to a specific product model, such as XX brand - monocrystalline 450W - bifacial module or YY model - fixed bracket system. The grid coordinates of the array layout are converted into actual layout drawing parameters including the number of rows, columns, vertical spacing, and horizontal spacing.

[0092] After parsing, the unit integration generates a complete product selection and layout strategy. This strategy explicitly specifies in the form of structured data: the specific product models and technical specifications used in each subarray, the row and column arrangement and installation tilt angle settings of the photovoltaic array, the front and rear row spacing of the support system and the width parameters of the maintenance passage, providing directly readable input conditions for the downstream automated structural calculation and compliance verification modules.

[0093] S2. Automated Structural Calculation and Compliance Verification Module: This module receives product selection and layout strategies, automatically calls structural calculation software for analysis and verification, and outputs structural compliance data and a draft bill of materials. Specifically, it includes:

[0094] S21. Computation Task Scheduling and Parameter Injection Unit: This unit acts as a bridge connecting AI intelligent decision-making and backend structural calculation software. It bears the core responsibility of data format conversion and automatic triggering of calculation tasks. It receives product selection and layout strategies and automatically calls external structural calculation software according to the strategies, injecting the required load and geometric parameters into the calculation model. Specifically:

[0095] S211. In the input receiving phase, the unit first obtains the product selection and layout strategy from the AI ​​intelligent decision-making and product selection module. This strategy is encapsulated in the form of structured data objects, including key fields such as component model code, array topology, installation tilt angle value, vertical and horizontal spacing matrix, and project geographical coordinates.

[0096] The unit's built-in strategy parsing engine initiates a multi-level parameter extraction process:

[0097] At the component level, the product model code is parsed and mapped to the corresponding physical attributes such as component size, weight, and wind load shape coefficient.

[0098] The array layer parses the layout pattern identifier, identifies it as a single-row, double-row, or multi-row layout scheme, and extracts geometric parameters such as the number of rows and columns, row spacing, and maintenance channel width.

[0099] The installation tilt angle value is analyzed at the tilt angle layer and converted into the actual angle configuration parameters of the support system.

[0100] After extraction, the unit stores the basic parameters into a temporary parameter pool and triggers the load data matching subprocess.

[0101] S212. Based on the project's geographical location information in the basic parameters, automatically match the corresponding wind load and snow load values ​​from the pre-set standard load library, or directly calculate the wind load and snow load values ​​using relevant specifications based on the project site's wind and snow data. Specifically:

[0102] A. During the load data matching phase, the unit performs intelligent retrieval from a pre-set standard load library based on the project's geographical location information in the basic parameter pool, accurate to the latitude and longitude coordinates of county-level administrative regions. The standard load library adopts a Geographic Information System (GIS) grid management system, dividing the national area into several climate zoning grids. Each grid node stores corresponding wind load and snow load data, including parameters such as 50-year return period basic wind pressure, ground roughness category, and wind pressure height variation coefficient, as well as parameters such as 50-year return period basic snow pressure, snow cover distribution coefficient, and roof snow unevenness distribution coefficient.

[0103] The element uses a spatial interpolation algorithm to map the project's geographical location to the nearest grid node and extracts the standard load data from that node. If the project is located at a grid boundary, the element uses bilinear interpolation to smooth the data, ensuring the accuracy and continuity of the load parameters. Furthermore, the element corrects the standard loads based on local correction factors such as array height and terrain. For example, a terrain correction coefficient is introduced for projects located on hillsides, and a wind vibration coefficient is introduced for support systems exceeding 20 meters in height.

[0104] B. Simultaneously, the unit supports a standardized direct calculation mode based on specific wind and snow data from the project site. If the project requirements provide measured or statistical wind and snow data such as wind speed, wind direction, snow depth, and snow density recorded by the local meteorological station, the unit will initiate the standardized direct calculation process:

[0105] a. For wind load calculations, the unit calculation shall be performed in accordance with the provisions of the "Code for Design of Building Structures". The specific process is as follows:

[0106] a1. Supports three methods for determining basic wind pressure:

[0107] Method 1: Directly query the basic wind speed or basic wind pressure of the corresponding region according to the appendix of the regulations or local standards of the project location;

[0108] Method 2: If weather station wind speed statistics are provided, then the standard formula shall apply. Calculate basic wind pressure ,in air density, This is the basic wind speed;

[0109] Method 3: If the customer explicitly specifies the wind speed or wind pressure value, then the customer-specified value will be used directly as the input.

[0110] If the return period, time interval, or unit of the provided wind speed or wind pressure is inconsistent with the requirements of the calculation specification, the unit will automatically convert it according to the conversion relationship or statistical method in the specification appendix to ensure that the input parameters meet the calculation requirements of the specification.

[0111] a2. Calculation of Wind Pressure Height Variation Coefficient: Based on the standard table, calculate the wind pressure height variation coefficient according to the ground roughness category of the project site and the ground clearance of the photovoltaic array. .

[0112] a3. Determination of wind load shape coefficient: Based on the array arrangement and installation tilt angle of the photovoltaic modules, the wind load shape coefficient is determined according to specifications or wind tunnel test data. .

[0113] a4. Application of Terrain Correction Factors: If the project is located in complex terrain such as mountainous areas or canyons, terrain correction factors should be introduced according to the specifications. .

[0114] a5. Wind vibration coefficient consideration: For flexible support systems with large height and low stiffness, the wind vibration coefficient should be calculated or selected according to the specifications. .

[0115] a6. Calculation of Standard Wind Load Value: Finally, the standard wind load value perpendicular to the surface of the photovoltaic module. According to the standard formula Calculate and generate the corresponding wind load values.

[0116] b. For snow load calculation, the unit calculation shall be performed in accordance with the "Code for Design of Building Structures" and relevant standards for photovoltaic brackets. The specific process is as follows:

[0117] Basic snow load determination: Based on the maximum snow load data (once in 50 years) provided by the local meteorological station, or snow depth and snow density conversion data, the basic snow load is determined directly or by interpolation according to the map in the appendix of the specification. .

[0118] Determination of roof snow distribution coefficient: Based on factors such as the installation tilt angle of the photovoltaic array and whether snow-blocking measures are considered, the roof snow distribution coefficient is determined in accordance with the specifications. .

[0119] Snow load standard value calculation: Finally, the standard value of snow load on the surface of the photovoltaic module. According to the standard formula Calculation. For multi-row arrays, the uneven distribution of snow between the front and rear rows also needs to be considered, and an uneven distribution coefficient is introduced.

[0120] After completing the direct calculations as described above, the unit records the calculated wind and snow load values, along with their calculation basis, as input for subsequent compliance verification. This mode complements the standard load library retrieval mode: when project requirements include reliable and specific wind and snow data, the direct calculation mode is prioritized to obtain more accurate loads that better reflect the actual climate conditions of the project; if no specific data is available, it automatically reverts to the standard load library matching mode to ensure the system's broad applicability.

[0121] S213. During the calculation input file generation stage, the element calls a pre-built calculation template library. This template library is custom-developed for mainstream structural calculation software, such as SAP2000, MIDAS, and ANSYS, and contains standardized file format templates that conform to the input specifications of each software. The element fills in the corresponding fields according to the template requirements with the geometric parameters in the foundation parameter pool, such as support cross-sectional dimensions, foundation depth, and array span, and the matched load data, such as nodal wind loads, line-distributed snow loads, and seismic actions, to generate a standardized calculation input file containing complete model definitions, material properties, cross-sectional characteristics, load cases, and analysis options.

[0122] During the file generation process, the unit performs syntax verification and logical consistency checks to ensure that the load combination meets the specifications, the boundary conditions are defined reasonably, and the analysis type is set correctly, thus avoiding calculation failures due to input errors.

[0123] S214. During the computation task submission phase, the unit achieves automated driving by calling the application programming interface (API) of the target structure computation software. For software that supports COM interfaces, the unit creates a software instance using an out-of-process call method, loads standardized computation input files, and starts batch computation mode; for software that provides a command-line interface, the unit constructs a command string containing parameters such as input file path, output file path, and solver options, and submits the computation task through a system process call.

[0124] After submission, the unit continuously monitors the computation process status, obtaining progress feedback through polling or event listening mechanisms. If computational anomalies are detected, such as model instability or convergence failure, the unit automatically records error logs and triggers a retry mechanism or manual intervention request. After computation is complete, the unit captures the mechanical analysis result file generated by the software and transfers it to the S22 unit for further processing.

[0125] S22. Automatic Compliance Verification Unit, connected to the computation task scheduling and parameter injection unit, serves as an automated review gate for structural safety, responsible for converting the numerical results of the structural calculation software into compliance judgments; specifically:

[0126] S221. During the result receiving phase, the unit acquires the mechanical analysis result file output by the structural calculation software. This file is usually in a common data format, such as a text file, XML file, or JSON file, and contains detailed mechanical response data of each structural component under various load conditions.

[0127] The unit's built-in result parsing engine initiates a multi-level data extraction process: At the component identification layer, key components, such as support columns, inclined beams, connection nodes, and foundations, are identified through component number, material type, and cross-sectional shape. At the response data layer, stress, strain, and displacement data of each component under ultimate limit state and normal serviceability limit state are extracted. Stress data includes axial stress, bending stress, shear stress, and combined stress; displacement data includes nodal horizontal displacement, vertical displacement, and angular displacement. The extracted data is indexed using component number and load case combination to facilitate subsequent comparison and retrieval.

[0128] S222. During the verification basis preparation phase, the unit automatically retrieves the corresponding mandatory standard limits and enterprise safety redundancy thresholds from its built-in specification library based on the type and material properties of key components. The specification library employs a hierarchical structure. The bottom layer stores basic parameters from national mandatory standards, such as material strength standard values, allowable slenderness ratios of components, and weld strength design values. The middle layer stores correction coefficients and adjustment parameters from industry recommended standards, such as temperature effect coefficients and fatigue calculation parameters specific to photovoltaic support structures. The top layer stores enterprise-defined safety redundancy thresholds, which introduce a safety margin coefficient of 1.1-1.3 times that of national standards, reflecting the stricter control requirements of enterprise internal standards compared to national standards. The unit uses component material grades and component categories as search keys to accurately locate the corresponding stress limits, displacement limits, and stability judgment criteria from the specification library.

[0129] S223. During the compliance assessment phase, the unit executes a detailed comparison process for each component and each working condition. For stress indicators, the maximum combined stress of each component is extracted and compared with the design value (actual stress / design strength) divided by the standard value of the material's yield strength. If the stress ratio is less than 1.0, the strength is deemed satisfactory; if the stress ratio is less than 0.8, it is marked as having a high safety margin. For displacement indicators, the maximum horizontal displacement of the extracted nodes is compared with 1 / 250 of the support height. If the displacement ratio is less than 1.0, the stiffness is deemed satisfactory; if the displacement ratio is less than 0.6, it is marked as having a low deformation risk. For stability indicators, the slenderness ratio of the component is calculated and compared with the allowable value in the specification, while the critical load for flexural and torsional buckling is verified. The comparison process adopts a veto system; if any component fails to meet any limit under any working condition, the entire component is deemed non-compliant. The unit records detailed comparison data for each component, including actual values, limit values, ratios, assessment results, and the specific specification clause numbers referenced, forming a traceable chain of assessment criteria.

[0130] S224. During the verification report generation phase, the unit integrates the judgment results of all components to generate a structured compliance verification report. The report adopts a hierarchical organization: the top layer is the overall compliance conclusion of the project, i.e., full compliance / partial compliance / non-compliance; the middle layer is a summary of the compliance status of each subarray or partition; and the bottom layer is a detailed data table and judgment log for each component. For component data judged to be compliant, the unit adds a compliance certification mark, such as a digital signature or hash value, and marks it as a mechanical analysis result that has passed the compliance judgment, which is then passed to the S23 unit for bill of materials generation. For components judged to be non-compliant, the unit highlights the non-compliant items in the report and points out possible optimization directions, such as increasing the cross-section, adjusting the spacing, or changing the material. At the same time, the information is fed back to the AI ​​intelligent decision-making module to trigger the iterative optimization process.

[0131] S23. Draft Bill of Materials Generation Unit, connected to the Automatic Compliance Verification Unit. This unit, as an automatic material data compilation module, is responsible for integrating compliant structural calculation results with product selection information into a draft bill of materials usable for the project; specifically:

[0132] S231. During the data receiving phase, the unit simultaneously acquires the mechanical analysis results from S22 (indicating compliance approval) and product information from S21 (product selection and layout strategy). The mechanical analysis results include detailed specifications for each component, such as the cross-sectional dimensions, length, and quantity of the support columns, and the type, dimensions, and concrete strength grade of the foundation. The product information includes procurement-level data such as the specific product models, component specifications, inverter models, and cable specifications used in each subarray. The unit establishes a data association mapping between component numbers and product model codes to ensure that each mechanical component can be traced back to its corresponding procured product item.

[0133] S232. During the bill of materials generation phase, the unit initiates a structured data processing workflow to convert the raw data into a standardized bill of materials format:

[0134] At the component entry level, the complete component technical specifications are retrieved from the enterprise's product database based on the product model. Parameters such as component type, peak power, open-circuit voltage, short-circuit current, size specifications, and weight are extracted and combined with the array layout to calculate the usage per set, forming component material entries.

[0135] At the support item level, based on the cross-sectional dimensions and length data in the mechanical analysis results, the standard profile specification codes are matched, such as hot-dip galvanized C-shaped steel and hot-dip galvanized angle steel, and the total usage of each specification is summarized to form support material items;

[0136] At the basic item level, based on the foundation type and size data, the concrete volume, steel reinforcement usage, and number of embedded parts are calculated and linked to the corresponding building material codes to form basic material items;

[0137] At the electrical item level, based on the number of components connected in series, string voltage and current parameters, and inverter input requirements, the specifications and quantity of DC cables, the specifications and quantity of AC cables, the number of combiner boxes, and the inverter capacity configuration are calculated to form electrical material items.

[0138] All items use a unified coding system, such as WBS coding or material coding, to facilitate subsequent integration with the ERP system.

[0139] S233. In the data organization phase, the unit generates a structured draft bill of materials. This draft is stored using a hierarchical data model. The top layer is the project overview, which includes summary indicators such as total installed capacity, total material cost, and total weight. The middle layer consists of sub-items, such as photovoltaic modules, support structures, electrical equipment, and civil engineering foundations. The bottom layer contains detailed items, including fields such as specific material codes, names, specifications, units, single-set usage, total usage, and material attributes.

[0140] The unit also includes pre-built data interfaces with mainstream engineering cost estimation software, allowing the initial draft of the bill of materials to be exported as XML or Excel format for cost engineers to conduct further price inquiries and cost calculations. Simultaneously, the initial draft includes data association pointers with the 3D parametric modeling module, ensuring that the S31 unit can directly read the specifications from the bill of materials when driving the modeling software, achieving seamless data flow integration. The final draft of the bill of materials output by the unit provides an accurate data foundation for downstream procurement, construction, and cost estimation processes.

[0141] S3. 3D Parametric Modeling and Drawing Module: This module receives structural compliance data and a draft bill of materials, drives 3D modeling software for automatic modeling, and outputs engineering drawings and an accurate bill of materials. Specifically, it includes:

[0142] S31. Parametric Model Driving and Update Unit: This unit serves as the core driver of the 3D modeling process, responsible for converting structurally compliant data into geometric and topological update instructions for the 3D model; specifically:

[0143] S311. During the data receiving phase, the unit acquires structural compliance data output by the automated structural calculation and compliance verification module. This data structure contains a complete set of structural parameters that have undergone compliance verification, typically encapsulated in XML or JSON format, covering the final design parameters of all structural components in the support system. The unit's built-in structural data parsing engine initiates a multi-dimensional parameter extraction process:

[0144] In the geometric attribute layer, the final cross-sectional dimension parameters of each component are parsed, such as the C-shaped steel cross-section specifications of the column and the rectangular tube cross-section dimensions of the inclined beam. These data are stored in string encoding form and need to be converted into a standard cross-section library index that can be recognized by 3D modeling software through regular expression matching and specification library mapping.

[0145] In the spatial positioning layer, the spatial coordinate parameters of key nodes of each component are analyzed. This coordinate system usually takes the center of the front row support foundation as the origin, the X-axis points to the horizontal arrangement direction of the array, the Y-axis points to the vertical arrangement direction, and the Z-axis is the vertical height direction. The coordinate values ​​are expressed as floating-point numbers in millimeters. The unit needs to perform coordinate system consistency verification to ensure alignment with the world coordinate system of the 3D modeling software.

[0146] In the connection layer, the connection angle parameters between components are analyzed, including the fixed connection angle between the column and the foundation, the hinged or rigid connection angle between the inclined beam and the column, and the overlap angle between the purlin and the inclined beam. These angle parameters are expressed as floating-point numbers in degrees and need to be converted into matching constraint values ​​in the 3D modeling software.

[0147] S312. During the model communication session establishment phase, the unit calls the corresponding software application interface based on the enterprise's pre-defined 3D modeling software type, such as SolidWorks, CATIA, Inventor, etc. For software supporting COM automation interfaces, the unit uses out-of-process late binding technology to create a software application instance, obtains the IDispatch pointer of the top-level assembly object, and then traverses to the root node of the pre-defined parametric 3D model; for software supporting .NET APIs, the unit instantiates an Application class object and opens a preset template file by adding a reference to the managed assembly provided by the software. The established communication session maintains a bidirectional data channel, supporting the unit to write parameters to the model and read status feedback from the model. The pre-built parametric 3D model is constructed using a top-down design paradigm. The top level is the overall assembly, the second level is the sub-array assemblies, and the bottom level is the support components. Each level of object exposes a standardized set of parameters. For example, the overall assembly contains global parameters such as installed capacity, number of array rows, and number of array columns. The sub-array assemblies contain layout parameters such as installation tilt angle, support height, and foundation spacing. The component layer contains component parameters such as cross-sectional specifications, material grade, and length. These parameters are marked as driving variables during model construction, and changes in their values ​​can trigger the automatic reconstruction of associated geometric features.

[0148] S313. During the parameter-driven update phase, the element will sequentially map the final cross-sectional dimensions, spatial coordinates, and connection angle parameters obtained from the analysis to the corresponding driving variables of the pre-set parametric 3D model through a communication session. The update process follows the principle of topological dependency order, prioritizing the updating of the bottom-level part parameters, then the mid-level assembly positioning parameters, and finally the top-level global parameters, ensuring the correctness of the geometric reconstruction.

[0149] For cross-sectional dimension parameters, the element calls the model's cross-sectional library interface through the API to query the predefined cross-sectional specification code, and points the cross-sectional feature attributes of the part to the specification definition to achieve automatic replacement of the cross-sectional profile;

[0150] For spatial coordinate parameters, the unit modifies the Mate mating relationship or Transform transformation matrix of the parts in the assembly through the API, and translates or rotates the parts from their original positions to new coordinate positions. For repeated parts arranged in an array, parametric array feature driving is adopted. Only the spacing parameter and the number of instances of the array parent feature need to be modified to achieve spatial repositioning of the entire row or column of parts.

[0151] For connection angle parameters, the unit modifies the angle constraint value in the mating relationship through the API, updates the direction vector of the hinge point or rigid node, and triggers the linkage update of related dependent features. For example, after the angle of the inclined beam changes, the lap surface of the purlin connected to it automatically adjusts the normal direction to maintain fit.

[0152] After parameter injection is complete, the unit sends a reconstruction command to the 3D modeling software via API. The software's internal parametric solver recalculates the geometric constraint equations, updates all features and entities that depend on these parameters, and generates a 3D model that perfectly matches the structural compliance data. During reconstruction, the unit monitors the model's reconstruction status, detects geometric conflicts or parameter out-of-bounds errors, and rolls back to the previous stable state and logs the error. If necessary, manual intervention for diagnosis is triggered.

[0153] S32. The automatic generation and annotation unit for engineering drawings is connected to the parametric model driving and updating unit. This unit, which is connected to the parametric model driving and updating unit, undertakes the key task of converting the 3D model into 2D drawings that conform to engineering drawing standards. Specifically:

[0154] S321. During the drawing generation preparation phase, the unit receives the updated 3D model from S31, which fully conforms to the geometric and topological definitions of the structural compliance data. The unit has a built-in predefined drawing rule library, which is customized based on enterprise drafting standards and industry specifications, defining the drawing expression requirements for different design stages. The rules are stored in XML script format and include subsets such as view configuration rules, annotation rules, and drawing frame template rules.

[0155] At the view configuration level, the rule defines that the main view should use front-view projection to show the overall elevation of the array, the top view should use top-view projection to show the array planar layout, and the side view should use left-view projection to show the side outline of the support and the details of the connection nodes. For complex nodes, local magnified views and sectional views also need to be generated.

[0156] At the scale setting level, the rules automatically match the optimal drawing scale based on the maximum outline length of the main parts in the current view to ensure clear expression of the part structure and optimal utilization of the drawing sheet. Specifically, the matching logic is as follows: when the maximum outline length of a part is ≤2 meters, a 1:20 scale is used to fully display connection details, welds, and holes; when the maximum outline length of a part is >2 meters and ≤8 meters, a 1:50 scale is used to balance the expression of the overall shape and local features; when the maximum outline length of a part is >8 meters, a 1:100 scale is used to ensure that large components are fully contained within the standard drawing sheet. The unit automatically measures the diagonal length of the bounding box of all parts in the current view via API, selects the maximum value as the scale matching basis, and supports automatic annotation of the selected scale on the drawing. When there are multiple parts with significant size differences in the same drawing, the unit uses a combination of magnified local views and the main view to ensure the readability of each part.

[0157] At the layer management level, rule-defined geometric entities, dimensions, text annotations, title blocks, etc., are placed in separate named layers for easy subsequent editing and printing control.

[0158] S322. During the automatic view generation stage, the unit extracts two-dimensional projection data from the three-dimensional model by calling the engineering drawing module API of the three-dimensional modeling software, based on predefined drawing rules.

[0159] The unit first creates a new engineering drawing document, setting the paper size to standard A1 or A0, and automatically selecting the orientation (horizontal or vertical) based on the array's aspect ratio. Then, the unit sequentially creates the front view, top view, and side view. During creation, the model projection direction and view placement are specified via API, and the software automatically generates a precise 2D projection outline after hidden lines are removed.

[0160] For repetitive components in the array, the unit employs an intelligent simplification strategy, drawing only the detailed structure of a standard support unit in its entirety in the main view, while the remaining units are represented by centerlines or simplified outlines. The arrangement pattern is explained in the top view through array annotations, avoiding information overload on the drawing. For critical connection nodes, the unit automatically identifies the node number and generates a magnified view. The magnification ratio is set according to the complexity of the node, typically 2:1 or 5:1, clearly showing details such as bolt connections and welding.

[0161] S323. During the automatic annotation stage, the unit starts the multi-dimensional annotation engine and automatically adds key dimensions, tolerances and part numbers to the view according to the drawing rules.

[0162] At the dimensioning level, the unit traverses the geometric features of the model through the API, extracts key dimensions such as length, angle, and spacing, and automatically selects the optimal dimensioning baseline. For example, the spacing of the support is dimensioned based on the center line of the foundation, and the height of the support is dimensioned based on the bottom surface of the column. The baseline dimensioning or continuous dimensioning method is used to keep the drawing neat. The dimensioning style follows the GB / T 4458.4 mechanical drawing standard, and the size of dimension lines, dimension extension lines, and arrows are uniformly set through the DimensionStyle object of the API.

[0163] At the tolerance annotation level, the unit automatically adds tolerance information based on the importance level of the component and the processing requirements. For example, H7 grade tolerance is marked for the critical connection hole diameter, and geometric tolerances such as perpendicularity and flatness are marked for welded parts. The tolerance values ​​are automatically matched from the built-in tolerance library according to the size segment.

[0164] At the part numbering level, the unit executes the assembly traversal algorithm to assign a unique number to each part and uses Balloon annotation objects to draw annotations on the view. At the same time, a part details column is automatically generated in the blank area of ​​the drawing, listing the part name, quantity, material, specifications and other information in numerical order.

[0165] After the annotation is completed, the unit performs collision detection to avoid overlap between dimension lines, text and geometric contours. If a conflict is detected, the layout is optimized by fine-tuning the annotation position or by using polylines to lead out the elements.

[0166] S324. During the drawing sheet refinement stage, the unit calls a pre-set standard drawing frame template via API. This template is stored in the enterprise drawing library as a block and includes standard fields such as design unit name, project name, drawing name, drawing number, scale, date, designer, checker, and approver. The unit automatically fills the drawing frame attribute fields with the metadata of the current drawing, namely the project name extracted from the requirements information in S11 and the drawing number rules extracted from the optimization strategy in S12, generating a title block that meets the requirements of document management. For complex projects that require multiple pages, the unit automatically adds a drawing index page, explaining the content and page number of each sub-drawing.

[0167] Finally, the unit saves the complete engineering drawings in both DWG and PDF formats. The DWG format retains complete layers and parametric information for subsequent editing and modification, while the PDF format is used for archiving, printing, and external delivery. Simultaneously, the unit generates a subset of assembly drawings and part machining drawings. The assembly drawings focus on expressing the assembly relationships and overall dimensions between components, while the part machining drawings generate separate drawings for each part requiring material cutting or machining, detailing machining dimensions, technical requirements, and material identifiers to directly guide workshop production.

[0168] S33. Material Information Synchronization and Bill of Materials Refinement Unit, connected to the Engineering Drawing Automatic Generation and Annotation Unit. This unit is responsible for refining and verifying the bill of materials from the initial draft to the precise version, ensuring the integrity and accuracy of material data; specifically:

[0169] S331. During the data receiving phase, the unit simultaneously receives the updated 3D model from S32 and the initial draft of the bill of materials from S23. The 3D model, as the authoritative source of geometric accuracy, has an assembly structure tree that fully records the instance information of all components, including component names, hierarchical relationships, geometric attributes, material properties, and quantity information. Each component exists as an independent node in the structure tree, and node attributes include pointers to part definitions, instance transformation matrices, and lists of mating relationships. The initial draft of the bill of materials serves as the initial basis for procurement and cost accounting, containing structured fields such as component type, specifications, unit usage, material information, and supplier codes. However, the initial draft data may contain specification deviations, quantity omissions, or incomplete information due to calculation simplifications or statistical errors in stages S21-S23.

[0170] S332. During the assembly structure tree traversal phase, the unit accesses the model's assembly manager object by calling the API of the 3D modeling software, and initiates a breadth-first or depth-first traversal algorithm to scan all component nodes in the assembly structure tree layer by layer.

[0171] For top-level assemblies, the unit extracts project-level metadata, such as the overall assembly name, design version number, and modification date; for subarray assemblies, it extracts information such as subarray identifier, installed capacity, and array number; for bottom-level parts, it extracts core attributes such as part number, part name, material grade, mass, volume, and geometric bounding box dimensions.

[0172] During the traversal, the unit uses an instance counting mechanism to count the number of references to the same part definition in different locations, accurately calculating the total number of instances for that part and avoiding errors from manual counting. For duplicate parts generated by parameterized arrays, the unit identifies the parent instance of the array features and the array parameters, using the formula Total Instances = Number of Rows × Number of Columns to ensure the accuracy of the count. Simultaneously, the unit extracts the geometric attributes of each part, such as the length of standard profiles, the unfolded dimensions of sheet metal parts, and the weld length of welded parts. This geometric data is obtained directly through the API's Measure tool or parameters.

[0173] S333. In the data comparison and difference identification stage, the unit automatically compares the material information extracted by traversing the assembly structure tree with the items in the initial draft of the bill of materials, using this information as baseline data. The comparison process employs a multi-key-value matching strategy, prioritizing the part number as the exact matching key to perform string-based exact matching between the part code in the 3D model and the material code in the initial draft. For items without a clear part number, a specification combination fuzzy matching is used, i.e., a combination of material grade + cross-section model + length range is used to search for similar items in the initial draft. The comparison dimensions cover key fields such as specification parameters, number of instances, and material properties.

[0174] At the specification level, the unit identified that the actual cross-sectional dimensions of a certain support column in the 3D model were C120×50×20×2.5, while the corresponding item in the initial draft was incorrectly written as C120×50×20×2.0, which was marked as a specification difference. At the quantity level, the unit calculated the total number of columns used through model statistics to be 480, while the initial draft recorded 450, which was marked as a quantity deviation. At the completeness level, the unit found that the 3D model contained lightning protection grounding flat iron parts, but there was no corresponding item in the initial draft, which was marked as a missing item. The difference identification algorithm established a difference matrix to record the matching status of each item in the initial draft, i.e., complete match, partial match, and no match, and generated a difference report.

[0175] S334. During the list refinement and correction phase, units automatically perform correction operations based on the comparison results:

[0176] For specification discrepancies, the unit uses precise specification parameters extracted from the 3D model to cover the erroneous parameters in the initial draft, and retains modification traces, such as adding revision marks and dates;

[0177] For quantity deviations, the total number of instances counted by the unit model is used to update the usage field in the initial draft, and the derived calculation fields such as total weight and total cost are also updated.

[0178] For missing items, the unit automatically inserts new entries in the initial draft. The entry data is completely extracted from the model attributes, including component name, specifications, material grade, unit weight, total weight, purchasing unit, etc., and the new entries are classified into the correct sub-items, such as lightning protection grounding flat iron being classified into the electrical equipment - grounding system category.

[0179] After the correction is completed, the unit performs a data consistency check to ensure that the hierarchical quantity relationship in the list is correct, such as the sum of the number of parts equals the number of their respective components, the sum of the number of each component equals the total usage, the total weight of materials is consistent with the quality statistics of the 3D model, and the material code conforms to the company's material coding rules.

[0180] Finally, the unit outputs a corrected and complete accurate bill of materials. This bill of materials is in a structured data format, supporting JSON, XML, or database table formats. It contains accurate material information for all components, and its data accuracy reaches the engineering application level that can be directly used for procurement ordering and construction material requisition, realizing closed-loop accurate control from design model to material data.

[0181] S4. One-stop solution package generation module, used to integrate product selection and layout strategies, structural compliance data, engineering drawings, and accurate bill of materials to automatically generate a complete design solution package; specifically including:

[0182] S41. Automatic Document Synthesis and Layout Unit: This unit receives four types of data: product selection strategy, structural compliance data, engineering drawings, and accurate bill of materials, and automatically generates technical documents according to predefined templates. Specifically:

[0183] The unit first parses and indexes each data source, establishing a cross-reference relationship between component numbers and document chapters.

[0184] Then, the template library is called to generate technical specifications: the project overview and equipment list are extracted from the strategy data and filled into the corresponding chapters, the design basis and structural conclusions are extracted from the compliance data to generate design specifications, and the equipment selection table is automatically inserted.

[0185] Synchronously generate structural calculation reports: extract calculation assumptions and load statistics from compliant data, extract component verification data from mechanical results to automatically generate verification details, highlight non-compliant items and provide qualitative evaluations.

[0186] After the document is generated, the unit calls the typesetting engine to uniformly set the font, line spacing, page margins and other formats, automatically builds a three-level directory and figure numbering, performs cross-reference validation, and finally outputs document types including .doc / .docx / .xls / .xlsx / .pdf, etc.

[0187] S42. Formatted Output and Packaging Unit, connected to the Automatic Document Synthesis and Layout Unit. This unit integrates technical documents, engineering drawings, and bills of materials, performs standardized format conversion, establishes logical connections, and packages them into a unified solution package. Specifically:

[0188] The unit first verifies the integrity of the three types of files, establishes hyperlinks between documents, such as links to bills of materials embedded in instruction manuals, and links to part drawings in calculation reports. The associated index is stored as a JSON file.

[0189] Subsequently, the formats were uniformly converted: technical documents were converted to PDF / A format, drawings were generated as PDFs and compatible DWG versions according to the printing style, and bills of materials were converted to Excel and CSV formats with formula protection. Files were standardized in naming according to the rule of "project number-specialty-type" and embedded with digital watermarks.

[0190] Finally, all files are packaged into a ZIP container, internally organized into three layers: 01 Technical Documents, 02 Engineering Drawings, and 03 Bill of Materials. This generates a solution package description document and metadata list containing a file list and checksums. The unit performs version control and quality access control checks, namely verifying the consistency of document information, the deviation of total material weight, and the validity of drawing numbers. Once these checks are passed, the unit is marked as deliverable.

[0191] S43. Distribution and System Interface Unit: Connects to the formatted output and packaging unit. This unit distributes the solution package to recipients such as owners, design institutes, and construction companies according to a predefined strategy, and synchronizes the bill of materials to the ERP system. Specifically:

[0192] The unit reads the recipient list and permission levels from the project requirements and matches the distribution strategy according to the project type: the owner receives the complete package, the contractor receives the drawings and bill of materials, and the supervisor receives editable versions of the documents and drawings. Distribution supports multiple methods such as email, FTP upload, and API push from the project management platform, monitoring the transmission status and recording logs. Recipients can view the status through a dashboard visualization. The unit synchronizes the accurate bill of materials through the ERP standard interface, automatically creating material master data or updating required quantities, generating purchase requisitions and triggering the approval process. The synchronization log records the results and error messages of each operation, supporting retry on failure. After distribution is completed, a delivery completion signal is written back to the project management platform, triggering subsequent construction processes and achieving seamless integration from design to procurement and construction.

[0193] Example 2

[0194] To verify the actual effect of the present invention, a simulation comparison experiment was designed and implemented to quantitatively evaluate its improvement in efficiency, quality, cost and consistency compared with the traditional manual design mode.

[0195] 1. Experimental Design

[0196] The experimental group used the AI ​​automatic design system provided by this invention; the control group used the traditional manual design process commonly used in the industry, that is, senior engineers used independent computing software, 3D CAD software and Office tools to manually design.

[0197] The experimental cases selected five representative real photovoltaic power station projects, covering two types of terrain: flat land and hilly land, as well as various support types such as fixed supports, tracking supports, flexible supports, BIPV / BAPV, to ensure the diversity of test scenarios.

[0198] The experimental environment was a controlled variable. Both groups of experiments used the same computer hardware configuration and followed the exact same customer input conditions, product selection range, and design specifications.

[0199] Evaluation metrics include: design efficiency, which is the total time required to complete a single project from input requirements to output of a complete and deliverable design solution package; design quality stability, which is the consistency and compliance of the output solution in terms of key performance indicators; economy, which is the total amount of steel used in the bill of materials (BOM) generated by the solution; and process compliance rate, which is the format standardization and completeness of various drawings, reports, and lists in the solution output.

[0200] 2. Experimental Procedure and Data Recording

[0201] For each project case, the experimental group and the control group carried out the design work in parallel, and the key node data were recorded as shown in Table 1 below:

[0202] Table 1: Comparison Results of Key Indicators in Simulation Experiments

[0203] 3. Results Analysis and Discussion

[0204] The experimental group had an average design time of 1.4 hours, which is more than 10 times more efficient than the control group's 12.9 hours. This directly verifies that the invention achieves significant time savings through the automatic generation of strategies by the AI ​​intelligent decision-making module and the unattended collaboration of the toolchain via automated structural calculation and compliance verification modules, as well as 3D parametric modeling and drawing modules, perfectly solving the industry pain point of low design efficiency.

[0205] The average maximum stress ratio of the key components in the experimental group was 0.83, which was slightly higher than that in the control group (0.79). This indicates that the AI ​​system made fuller use of material properties and reduced overly conservative design while meeting safety standards, which is consistent with the material saving results.

[0206] The variance of this indicator in the experimental group was only 0.014, far lower than the 0.138 in the control group, and the quality fluctuation was reduced by about 87%. This strongly demonstrates that the present invention, by encoding and solidifying design rules and expert experience into the system, completely eliminates the quality fluctuation caused by differences in state and experience in manual design, and achieves a high degree of standardization and stability of output.

[0207] The experimental group's average steel consumption was reduced by 6.2% compared to the control group. This is directly attributable to the global multi-objective optimization capability of the AI ​​intelligent decision-making and product selection module, which can find a better material configuration solution under multiple constraints such as cost and safety, thus verifying its cost reduction potential.

[0208] The experimental group achieved a 100% compliance rate for complete output documents in all cases, while the control group averaged only 88.4%. This demonstrates the effectiveness of the one-stop solution package generation module and the automated verification mechanism throughout the entire process, ensuring the professionalism and reliability of the deliverables and eliminating human error.

[0209] 4. Experimental Conclusions

[0210] This simulation experiment, through rigorous comparative verification, fully demonstrates that the present invention achieves a leapfrog improvement in design efficiency, quality stability, economy, and process standardization compared to the traditional manual design mode. This system successfully makes tacit knowledge explicit, integrates discrete tools, and automates manual processes, effectively solving all the core pain points mentioned in the background technology. Furthermore, it maximizes design value through a data-driven approach, providing a revolutionary solution for the field of photovoltaic support system design.

[0211] As can be seen from the above description, the embodiments of the present invention achieve the following technical effects:

[0212] This system achieves the knowledge-based encapsulation of design experience and the global optimization of decision-making capabilities, fundamentally solving the core pain points of design quality relying on personal experience, difficulty in inheritance, and insufficient optimization. By constructing an AI-powered intelligent decision-making and product selection module, it encodes the company's accumulated, hard-to-transmit expert experience and massive historical successful solutions into a computable, callable, and iterative dedicated optimization algorithm and structured knowledge base, creating an AI design expert unaffected by personnel turnover. Based on specific project requirements, this expert system can dynamically and globally optimize under multiple constraints such as cost, material usage, and structural safety, outputting optimal product selection and layout strategies that far surpass the judgment of local human experience. This innovation directly brings about a qualitative change in design quality, from large fluctuations to continuous stability, and achieves an average direct material cost saving of 5%-8% due to global optimization. Simultaneously, it securely transfers core design knowledge from engineers' minds to the enterprise system, building a technological moat that is difficult to replicate.

[0213] Overcoming the challenges of automated collaboration and 100% compliance verification across multi-source heterogeneous toolchains, this invention revolutionizes the entire design process and achieves zero-risk quality control. By constructing an automated structural calculation and compliance verification module, it achieves, for the first time, deep, automated, and real-time collaboration between an AI decision-making system and external professional engineering software as well as internal enterprise systems. This system automatically analyzes design strategies, invokes professional software to perform calculations, and extracts results for full-scale automated compliance comparison. It compresses the previously fragmented, time-consuming 8-16 hour manual process into a fully automated pipeline, reducing the design time for a single solution to 1-2 hours, improving efficiency by approximately 10 times. More importantly, through automatic verification using a built-in specification library, it reduces design compliance risks caused by human error to near zero, achieving absolute reliability and standardized output of design results, providing technical support for rapid market response and the expansion of time-sensitive businesses.

[0214] This invention constructs a complete digital closed loop from intelligent decision-making to business implementation, significantly improving enterprise operational efficiency, supply chain collaboration, and creating new business model possibilities. Through a one-stop solution package generation module and the linkage of the aforementioned modules, this invention achieves automated delivery from demand input to a complete solution package, automatically synchronizing the quotation list to the quotation system and the accurate bill of materials (BOM) to the ERP system, thus connecting the key links of design data-driven production and supply chain. This not only significantly shortens the quotation and delivery cycle and enhances market competitiveness but also enables predictive collaboration with upstream suppliers through accurate BOMs, optimizing inventory and reducing capital occupation. Ultimately, the system not only serves as an internal productivity tool but can also be exported to the industry as a high-value-added software product or service, opening up a second growth curve for enterprises and achieving a leap from improving internal efficiency to empowering the external industrial chain.

[0215] The embodiments and / or implementation methods described above are merely preferred embodiments and / or implementation methods for implementing the technology of the present invention, and are not intended to limit the implementation methods of the technology of the present invention in any way. Any person skilled in the art can make some modifications or alterations to other equivalent embodiments without departing from the scope of the technical means disclosed in the present invention, but these should still be regarded as the technology or embodiments that are substantially the same as the present invention.

[0216] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. The above descriptions are only preferred embodiments of this application. It should be noted that due to the limitations of written expression, while there are objectively infinite specific structures, those skilled in the art can make several improvements, modifications, or changes without departing from the principles of this application, and can also combine the above technical features in an appropriate manner. These improvements, modifications, changes, or combinations, or the direct application of the inventive concept and technical solution to other situations without modification, should all be considered within the scope of protection of this application.

Claims

1. An AI-powered automatic design system, characterized in that, include: The AI-powered intelligent decision-making and product selection module is used to output product selection and layout strategies based on project requirements information, by calling dedicated optimization algorithms and knowledge bases. The automated structural calculation and compliance verification module is used to receive the product selection and layout strategy, automatically call the structural calculation software for analysis and verification, and output structural compliance data and a draft bill of materials. The 3D parametric modeling and drawing module is used to receive the structural compliance data and the initial draft of the bill of materials, drive the 3D modeling software to automatically model, and output engineering drawings and an accurate bill of materials. The one-stop solution package generation module is used to integrate the product selection and layout strategies, structural compliance data, engineering drawings and accurate bill of materials to automatically generate a complete design solution package.

2. The AI ​​automatic design system according to claim 1, characterized in that, The AI-powered intelligent decision-making and product selection module specifically includes: The information receiving and parsing unit is used to receive project requirement information, and to perform data cleaning and standardization processing on the project requirement information, converting it into a unified structured data format that the system can process. The optimization decision-making unit, connected to the information receiving and parsing unit, is used to call a dedicated optimization algorithm and knowledge base for matching and calculation based on the structured project requirement information. The strategy generation unit, connected to the optimization decision unit, is used to analyze the calculation results of the optimization decision unit and integrate them to generate the product selection and layout strategy, which includes specific product models, array layouts, installation tilt angles, and spacing parameters.

3. The AI ​​automatic design system according to claim 2, characterized in that, The specific workflow of the optimization decision-making unit is as follows: Receive project requirement information in a unified structured data format output by the information receiving and parsing unit; Based on the key features in the project requirements information, retrieve matching historical design schemes and applicable expert experience rules from the knowledge base; The dedicated optimization algorithm is invoked, with the retrieved historical design schemes as the initial solution and the applicable expert experience rules as constraints, to construct and solve a multi-objective optimization model oriented towards cost, material usage, and structural safety. The solution results of the multi-objective optimization model are output as the decision-making basis for generating the product selection and layout strategy.

4. The AI ​​automatic design system according to claim 1, characterized in that, The automated structural calculation and compliance verification module specifically includes: The task scheduling and parameter injection unit is used to receive the product selection and layout strategy, and automatically call the external structural calculation software according to the strategy to inject the load parameters and geometric parameters required for calculation into the calculation model. The automatic compliance verification unit is connected to the calculation task scheduling and parameter injection unit. It is used to obtain the mechanical analysis results output by the structural calculation software and automatically compare and judge the compliance of the results according to the built-in design specifications. The bill of materials draft generation unit is connected to the specification compliance automatic verification unit. It is used to generate structured bill of materials data, including component type, specifications, single set usage and material information, based on the mechanical data of the components that have passed the compliance verification and the corresponding product selection information.

5. The AI ​​automatic design system according to claim 4, characterized in that, The specific workflow of the computation task scheduling and parameter injection unit is as follows: The system receives product selection and layout strategies from the AI ​​intelligent decision-making and product selection module, analyzes the product selection and layout strategies, and extracts product model, installation tilt angle, array spacing and arrangement parameters as basic parameters for calculation input. Based on the project's geographical location information in the basic parameters, the corresponding wind load and snow load values ​​are automatically matched from the pre-set standard load library, or the wind load and snow load values ​​are directly calculated using relevant specifications based on the wind and snow data of the project site. Integrate the basic parameters with the matched or calculated load data, and call the preset calculation template to generate a standardized calculation input file that conforms to the input format of the target structure calculation software; The application programming interface of the target structure calculation software is invoked, and the standardized calculation input file is submitted to start automated structure calculation.

6. The AI ​​automatic design system according to claim 5, characterized in that, The specific workflow of the automatic compliance verification unit is as follows: Receive the mechanical analysis results output by the structural calculation software, parse the mechanical analysis results, and extract the stress, strain, and displacement data of key components; Based on the type and material properties of the key components, the corresponding mandatory standard limits and enterprise safety redundancy thresholds are automatically retrieved from the built-in specification library. The extracted stress, strain, and displacement data are compared one by one with the mandatory standard limits and the enterprise safety redundancy threshold, and the compliance of each component with the specifications is determined based on the comparison results. The output includes a compliance verification report containing the compliance status of all components, specific data, and the basis for judgment. Component data that are judged to comply with the specifications in the report are marked as mechanical analysis results that have passed the compliance judgment.

7. The AI ​​automatic design system according to claim 1, characterized in that, The 3D parametric modeling and mapping module specifically includes: The parametric model driving and updating unit is used to receive the structural compliance data and automatically drive the parametric model in the external 3D modeling software to update and reconstruct it based on the geometric parameters and constraints therein. The automatic engineering drawing generation and annotation unit is connected to the parametric model driving and updating unit. It is used to automatically generate the front view, top view and side view from the updated 3D model according to the predefined drawing rules, and automatically annotate key dimensions, tolerances and part numbers. Finally, it inserts a standard drawing frame to generate engineering drawings, assembly drawings and part processing drawings that conform to engineering drawing specifications. The material information synchronization and bill of materials refinement unit is connected to the engineering drawing automatic generation and annotation unit. It is used to automatically count and extract the material information of all components from the final three-dimensional model, update and refine the initial draft of the bill of materials, and output the accurate bill of materials.

8. The AI ​​automatic design system according to claim 7, characterized in that, The specific workflow of the parameterized model-driven and update unit is as follows: Receive structural compliance data output by the automated structural calculation and compliance verification module, parse the structural compliance data, and extract the final cross-sectional dimensions, spatial coordinates, and connection angle parameters between the components for each structural component defined therein; Call the application programming interface of the 3D modeling software to create a communication session with the preset parametric 3D model. Based on the final cross-sectional dimensions, spatial coordinates and connection angle parameters obtained from the analysis, modify the values ​​of the driving variables of the corresponding features in the preset parametric 3D model in sequence through the communication session. The 3D modeling software is triggered to perform a model reconstruction operation based on the updated driving variable values, generating an updated 3D model that perfectly matches the structural compliance data.

9. The AI ​​automatic design system according to claim 8, characterized in that, The specific workflow of the material information synchronization and list refinement unit is as follows: Receive the updated 3D model from the automatic engineering drawing generation and annotation unit and the initial draft of the bill of materials from the automated structural calculation and compliance verification module; Traverse the assembly structure tree of the updated 3D model, and for each component, extract its geometric properties, predefined material properties, and number of instances; The extracted geometric and material properties are automatically compared with the corresponding items in the initial draft of the bill of materials to identify specification differences, quantity deviations, or missing items. Based on the comparison results, the specifications and parameters in the initial draft of the bill of materials are automatically corrected, the quantities are updated, and missing component information is supplemented. The corrected and supplemented bill of materials is then output as a structured and accurate bill of materials containing precise material information for all components.

10. The AI ​​automatic design system according to claim 1, characterized in that, The one-stop solution package generation module specifically includes: The document automatic synthesis and typesetting unit is used to receive and integrate the product selection and layout strategy, structural compliance data, engineering drawings and accurate bill of materials, and automatically generate technical document parts including technical specifications and structural calculation reports according to predefined document templates; The formatted output and packaging unit, connected to the document automatic synthesis and typesetting unit, is used to standardize and logically associate the technical document, engineering drawings and precise bill of materials, and package them into a unified, deliverable complete design solution package. The distribution and system interface unit is connected to the formatted output and packaging unit, and is used to automatically distribute the complete design package to predefined recipients and synchronize the accurate bill of materials to the enterprise resource planning system.

Images