A prefabricated product air pipe intelligent processing system and method based on digital modeling

Abstract

The application relates to the technical field of intelligent processing, and discloses a prefabricated product air pipe intelligent processing system and method based on digital modeling. The system comprises a parameter acquisition module, a feature extraction module, a pressure drop calculation module, a parameter optimization module, a process mapping module and a self-adaptive optimization module. The layout parameters, performance requirements and physical geometric attributes of the prefabricated product air pipe are acquired. The physical geometric attributes are filtered and denoised, and structured coding is performed to obtain a multi-modal feature vector. According to the consistency constraint of fluid mechanics, the feature vector is forward propagated to obtain a pressure drop gradient distribution. Based on the pressure drop gradient distribution, the feature vector is iteratively adjusted to obtain target manufacturing guide parameters. The target manufacturing guide parameters are mapped to a material-process mapping knowledge base to obtain a numerical control processing instruction set. The instruction set is executed, real-time processing data is acquired, and the material-process mapping knowledge base is optimized and updated. The application can improve the efficiency of air pipe processing.

Description

Technical Field

[0001] This invention relates to the field of intelligent manufacturing technology, and in particular to an intelligent manufacturing system and method for prefabricated air ducts based on digital modeling. Background Technology

[0002] Prefabricated ducts are widely used in engineering fields such as ventilation and air conditioning, and industrial ventilation. They are a core component of fluid transmission systems. As engineering complexity increases, the requirements for the precision, performance matching, and industrialization efficiency of duct processing continue to rise. Digital modeling and intelligent processing technologies are gradually becoming the development direction of the duct processing field. By using data-driven approaches to achieve synergy between design parameters and processing execution, duct processing is being transformed towards precision and efficiency.

[0003] Existing technologies suffer from two core technical problems: First, the integration of layout parameters and fluid dynamics performance requirements lacks a standardized processing procedure, and the correlation analysis between path topology data and fluid constraints is inaccurate, leading to significant deviations in the calculation of key performance parameters such as pressure drop, making it difficult to ensure that the actual performance of the ductwork matches the design requirements. Second, the mapping of geometric performance parameters from the design stage to CNC machining instructions lacks accurate material-process knowledge base support, and the machining process lacks real-time data feedback and knowledge base iteration mechanisms, resulting in a disconnect between design and production, and insufficient machining efficiency and quality stability. Therefore, how to improve the efficiency of ductwork machining has become an urgent problem to be solved. Summary of the Invention

[0004] This invention provides an intelligent processing system and method for prefabricated air ducts based on digital modeling, in order to solve the problems mentioned in the background art.

[0005] To achieve the above objectives, the present invention provides an intelligent processing system for prefabricated ductwork based on digital modeling, characterized in that the system includes a parameter acquisition module, a feature extraction module, a pressure drop calculation module, a parameter optimization module, a process mapping module, and an adaptive optimization module, wherein: The parameter acquisition module is used to acquire the layout parameters and performance requirements of the prefabricated air duct, and obtain the physical geometric properties of the prefabricated air duct. The feature extraction module is used to filter the physical geometric attributes and perform structured encoding on the denoised physical geometric attributes to obtain the multimodal feature vector of the physical geometric attributes. The pressure drop calculation module is used to perform forward propagation of the multimodal feature vector according to the consistency constraints in fluid mechanics to obtain the pressure drop gradient distribution of the prefabricated air duct. The parameter optimization module is used to iteratively adjust the geometric performance characteristic parameters in the multimodal feature vector based on the backpropagation optimization of the pressure drop gradient distribution, so as to obtain the target manufacturing guidance parameters of the prefabricated air duct. The process mapping module is used to map the target manufacturing guidance parameters to the material-process mapping knowledge base to obtain the CNC machining instruction set of the prefabricated air duct; The adaptive optimization module is used to execute the CNC machining instruction set, obtain real-time data during the duct processing, and optimize and update the material-process mapping knowledge base based on the real-time data.

[0006] In a preferred embodiment, when the parameter acquisition module acquires the layout parameters and performance requirements of the prefabricated ductwork and obtains the physical geometric properties of the prefabricated ductwork, it is specifically used for: Obtain the layout parameters and performance requirements of prefabricated air ducts; The layout parameters of the prefabricated air duct are digitally analyzed to obtain the path topology data of the prefabricated air duct. The performance requirements are extracted in a structured manner to obtain the fluid dynamic constraints of the prefabricated air duct; Based on the fluid dynamics constraints, a geometric correlation analysis is performed on the path topology data to obtain the initial geometric properties of the prefabricated duct. The initial geometric properties are normalized to obtain the physical geometric properties of the prefabricated duct.

[0007] In a preferred embodiment, when the feature extraction module performs filtering on the physical geometric attributes and structured encoding on the denoised physical geometric attributes to obtain the multimodal feature vector of the physical geometric attributes, it is specifically used for: The physical geometric attributes are subjected to time-frequency domain filtering to obtain the duct geometric feature sequence of the physical geometric attributes; Multi-scale feature fusion is performed on the geometric feature sequence of the duct to obtain a multi-scale feature representation of the prefabricated duct. The multi-scale feature representation is encoded using a graph neural network to obtain the spatial topological feature tensor of the prefabricated duct. Extract multiple topological feature vectors from the spatial topological feature tensor, and combine the multiple topological feature vectors into a topological feature vector group; The topological feature vector group is embedded in a high-dimensional manifold to obtain the multimodal feature vector of the physical geometric attribute.

[0008] In a preferred embodiment, when the pressure drop calculation module performs forward propagation of the multimodal feature vectors based on consistency constraints in fluid mechanics to obtain the pressure drop gradient distribution of the prefabricated duct, it is specifically used for: The multimodal feature vectors are spatially discretized to obtain the discretized spatial data of the prefabricated air duct. The flow field characteristics are evolved from the discretized spatial data to obtain the internal velocity field distribution of the prefabricated duct.

[0009] In a preferred embodiment, the voltage drop calculation module is further configured to: Based on the preset wall normal gradient distribution, the field characteristics of the internal velocity field distribution are analyzed to obtain the velocity gradient distribution of the prefabricated air duct. The formula for calculating the pressure drop gradient distribution is as follows: ; In the formula, The pressure drop gradient distribution is... The total number of discrete segments in the spatial topological feature tensor. The first one obtained based on the physical geometric properties The cross-sectional area of ​​each discrete segment The first one defined based on the path topology data The spatial area occupied by each discrete section of the air duct. For boundary operators, denote the spatial region after the boundary operator. Take the boundary. For the first The wall boundaries of each discrete segment are determined based on the path topology data. The wall shear stress is obtained by converting the viscous stress relationship based on the velocity gradient distribution. Let the area of ​​the infinitesimal element on the wall boundary be . The first one obtained based on the path topology data The path element vector of each discrete segment Indicates the first The surface integral of the shear stress on a discrete segment of the wall. This represents the cumulative summation along the path for all discrete segments based on the path topology data; The velocity gradient distribution is transformed into a viscous stress relationship to obtain the shear stress distribution of the velocity gradient distribution; The pressure drop gradient distribution of the prefabricated duct is obtained by performing friction resistance accumulation processing on the shear stress distribution.

[0010] In a preferred embodiment, when the parameter optimization module performs backpropagation optimization based on the pressure drop gradient distribution to iteratively adjust the geometric performance characteristic parameters in the multimodal feature vector to obtain the target manufacturing guidance parameters of the prefabricated duct, it is specifically used for: By performing an adjoint differential sensitivity analysis on the pressure drop gradient distribution, the adjoint gradient field of the pressure drop gradient distribution is obtained. Based on the accompanying gradient field, the geometric performance characteristic parameters are optimized by backpropagation to obtain the geometric parameter correction amount of the prefabricated air duct. The formula for calculating the geometric parameter correction is as follows: ; = , ; In the formula, and The geometric performance characteristic parameters are respectively for the t-th and t+1-th iterations. The preset learning rate, Let be the sensitivity coefficient at the t-th iteration. This is the cumulative momentum of the sensitivity coefficient. The adaptive adjustment amount of the sensitivity coefficient is denoted as , where and To estimate the attenuation rate using moments, It is the numerical stability constant; The geometric parameter correction amount is optimized by adaptive moment estimation to obtain the parameter update value of the geometric parameter correction amount; By projecting the updated parameter values ​​onto the manufacturing process constraints, feasible parameter solutions for the updated parameter values ​​are obtained.

[0011] In a preferred embodiment, the parameter optimization module is further configured to: The feasible parameters are demapped to the convergence solution set defined by the convergence criterion to obtain the optimized parameters of the prefabricated air duct. The optimization parameters are subjected to Pareto screening to obtain a set of candidate parameters for the optimization parameters; A manufacturing feasibility assessment is performed on the candidate parameter set to obtain the target manufacturing guidance parameters for the prefabricated air duct.

[0012] In a preferred embodiment, when the process mapping module executes the mapping of the target manufacturing guidance parameters to the material-process mapping knowledge base to obtain the CNC machining instruction set of the prefabricated air duct, it is specifically used for: The target manufacturing guidance parameters are decomposed to obtain the parameter feature components of the prefabricated air duct; Based on a pre-set material-process knowledge base, multi-dimensional constraint matching is performed on the parameter feature components to obtain an initial process scheme for the parameter feature components. The processing paths and parameters in the initial process scheme are optimized collaboratively to obtain the optimized process configuration of the prefabricated air duct; Numerical control instructions are generated for the optimized process configuration to obtain the numerical control machining instruction set for the prefabricated air duct.

[0013] In a preferred embodiment, when the adaptive optimization module executes the CNC machining instruction set to obtain real-time data during the duct machining process, and optimizes and updates the material-process mapping knowledge base based on the real-time data, it is specifically used for: The CNC machining instruction set is digitally processed to obtain real-time processing status data of the prefabricated air duct; Multi-source data fusion analysis is performed on the real-time processing status data to obtain the quality characteristic parameters of the prefabricated air duct during the processing. Key quality indicators are extracted from the quality characteristic parameters and their matching degree is evaluated with the material-process mapping knowledge base to obtain the optimization requirement indicators of the material-process mapping knowledge base. The knowledge base is incrementally updated based on the optimization requirement indicators to obtain the optimized material-process mapping knowledge base for the prefabricated air duct.

[0014] To address the above problems, the present invention also provides a method for intelligent processing of prefabricated air ducts based on digital modeling, the method comprising: S1: Used to obtain the layout parameters and performance requirements of prefabricated air ducts, and to obtain the physical geometric properties of prefabricated air ducts; S2: Used to filter the physical geometric attributes and perform structured encoding on the denoised physical geometric attributes to obtain the multimodal feature vector of the physical geometric attributes; S3: Used to perform forward propagation of the multimodal feature vector according to the consistency constraints in fluid mechanics to obtain the pressure drop gradient distribution of the prefabricated air duct; S4: Used for backpropagation optimization based on the pressure drop gradient distribution to iteratively adjust the geometric performance characteristic parameters in the multimodal feature vector to obtain the target manufacturing guidance parameters of the prefabricated duct. S5: Used to map the target manufacturing guidance parameters to the material-process mapping knowledge base to obtain the CNC machining instruction set of the prefabricated air duct; S6: Used to execute the CNC machining instruction set, obtain real-time data during the duct processing, and optimize and update the material-process mapping knowledge base based on the real-time data.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention collects data from multiple channels through a parameter acquisition module and combines it with industry standards to transform layout information into structured topology data and performance requirements into standardized constraints. Then, the feature extraction module denoises and encodes the data to form multimodal feature vectors. Finally, the pressure drop calculation module calculates the pressure drop gradient distribution based on fluid dynamics consistency constraints and accurate formulas, which greatly reduces the deviation in performance calculation.

[0016] 2. This invention optimizes geometric parameters and assesses manufacturing feasibility by combining pressure drop data with a parameter optimization module. The process mapping module generates standardized CNC instructions based on a material-process knowledge base. The adaptive optimization module collects real-time processing data and incrementally updates the knowledge base, forming a closed loop of "design-processing-feedback-iteration" to ensure that the parameters are machinable and the process is highly adaptable. Attached Figure Description

[0017] Figure 1 This is a system architecture diagram of a prefabricated air duct intelligent processing system based on digital modeling, provided in an embodiment of the present invention. Figure 2 This is a flowchart illustrating an intelligent processing method for prefabricated air ducts based on digital modeling, provided as an embodiment of the present invention.

[0018] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, 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 belong to some, but not all, embodiments of the present invention. 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.

[0020] The terminology used in the embodiments of this invention is for the purpose of describing particular embodiments only and is not intended to limit the invention. The singular forms “said” and “the” as used in the embodiments of this invention and the appended claims are also intended to include the plural forms, and “multiple” generally includes at least two unless the context clearly indicates otherwise.

[0021] Depending on the context, the word "if" or "if" as used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrase "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0022] Furthermore, the timing of the steps in the following method embodiments is merely an example and not a strict limitation.

[0023] In practice, the server-side equipment deployed in the prefabricated duct intelligent processing system based on digital modeling may consist of one or more devices. This system can be implemented as a business instance, a virtual machine, or hardware devices. For example, it can be implemented as a business instance deployed on one or more devices in a cloud node. Simply put, it can be understood as software deployed on a cloud node, providing the prefabricated duct intelligent processing system based on digital modeling to various user terminals. Alternatively, it can be implemented as a virtual machine deployed on one or more devices in a cloud node, with application software installed to manage various user terminals. Or, it can also be implemented as a server-side system composed of numerous identical or different types of hardware devices, with one or more devices configured to provide the prefabricated duct intelligent processing system based on digital modeling to various user terminals.

[0024] In terms of implementation, the intelligent processing system for prefabricated ductwork based on digital modeling and the user terminal are mutually compatible. That is, if the intelligent processing system for prefabricated ductwork based on digital modeling is implemented as an application installed on a cloud service platform, then the user terminal is a client that establishes a communication connection with the application; or if the intelligent processing system for prefabricated ductwork based on digital modeling is implemented as a website, then the user terminal is implemented as a webpage; or if the intelligent processing system for prefabricated ductwork based on digital modeling is implemented as a cloud service platform, then the user terminal is implemented as a mini-program in an instant messaging application.

[0025] like Figure 1 The figure shown is a system architecture diagram of a prefabricated air duct intelligent processing system based on digital modeling provided in an embodiment of the present invention.

[0026] The intelligent processing system 100 for prefabricated ductwork based on digital modeling described in this invention can be located on a cloud server. In terms of implementation, it can function as one or more service devices, or as an application installed on the cloud (e.g., a mobile service operator's server, server cluster, etc.), or it can be developed into a website. Depending on the functions implemented, the intelligent processing system 100 for prefabricated ductwork based on digital modeling may include a parameter acquisition module 101, a feature extraction module 102, a pressure drop calculation module 103, a parameter optimization module 104, a process mapping module 105, and an adaptive optimization module 106. The modules described in this invention can also be called units, referring to a series of computer program segments that can be executed by an electronic device's processor and perform a fixed function, stored in the electronic device's memory.

[0027] In this embodiment of the invention, in the intelligent processing system for prefabricated ductwork based on digital modeling, each of the above modules can be implemented independently and can be invoked by other modules. Invocation here can be understood as a module connecting to multiple modules of another type and providing corresponding services to those connected modules. In the intelligent processing system for prefabricated ductwork based on digital modeling provided by this embodiment of the invention, the applicable scope of the system architecture can be adjusted by adding modules and directly invoking them without modifying the program code, achieving cluster-based horizontal expansion to quickly and flexibly expand the intelligent processing system for prefabricated ductwork based on digital modeling. In practical applications, the above modules can be set in the same device or different devices, or they can be set in a virtual device, such as a service instance in a cloud server.

[0028] The following describes, with reference to specific embodiments, each component and specific workflow of the intelligent processing system for prefabricated air ducts based on digital modeling: The parameter acquisition module 101 is used to acquire the layout parameters and performance requirements of the prefabricated air duct, and obtain the physical geometric properties of the prefabricated air duct. In this embodiment of the invention, when the parameter acquisition module obtains the layout parameters and performance requirements of the prefabricated duct and acquires the physical geometric properties of the prefabricated duct, it is specifically used for: Obtain the layout parameters and performance requirements of prefabricated air ducts; The layout parameters of the prefabricated air duct are digitally analyzed to obtain the path topology data of the prefabricated air duct. The performance requirements are extracted in a structured manner to obtain the fluid dynamic constraints of the prefabricated air duct; Based on the fluid dynamics constraints, a geometric correlation analysis is performed on the path topology data to obtain the initial geometric properties of the prefabricated duct. The initial geometric properties are normalized to obtain the physical geometric properties of the prefabricated duct.

[0029] The module connects to 3D modeling tools such as AutoCAD and Revit through a two-way data interaction interface, automatically reading the 3D model data of the duct layout. It also provides a visual manual input window for users to supplement layout details not reflected in the model. In terms of performance requirements, the module can parse the technical specifications uploaded by users and extract indicators such as design air volume and allowable air pressure loss. It also loads the fluid transmission performance benchmark in GB50243 "Code for Acceptance of Construction Quality of Ventilation and Air Conditioning Engineering" by default, ensuring that the original data covers the core information of layout and performance. The 3D model data is converted into a Cartesian coordinate system, and key geometric nodes of the duct are located and their 3D coordinates are recorded using feature recognition technology. Then, the duct is divided into segments based on these nodes, and the length, direction, cross-sectional shape, and dimensions of each segment are calculated. Finally, the data is stored in a structured format of "node table - segment table - connection relationship table," forming path topology data that can be directly accessed by subsequent modules. Extract fluid-related keywords and corresponding numerical units such as "airflow," "pressure loss," and "velocity" from the raw performance requirement data. Based on fluid calculation needs, the extracted information is categorized into four main types: flow constraints, pressure drop constraints, velocity constraints, and media property constraints. Each type of constraint is stored in the format of "constraint type - constraint value - allowable deviation - priority," forming standardized fluid dynamics constraints. First, based on the correlation between fluid velocity, flow rate and pipe cross-sectional area, the minimum threshold of pipe cross-sectional area that meets the velocity requirement is derived; then, each pipe segment is traversed to check whether parameters such as cross-sectional dimensions, length, and bending angle meet the constraint threshold; pipe segments that do not meet the threshold are initially adjusted, and finally, all pipe segments are integrated after verification and adjustment to form the initial geometric attributes that completely describe the geometric characteristics of the duct. The initial geometric attribute parameters are standardized in terms of dimensions. Then, different types of geometric parameters, such as length, cross-sectional area, and angle, are processed separately. For example, the length parameter is mapped to the interval [0,1]. By calculating the ratio of the difference between the original length of the pipe segment and the extreme lengths of all pipe segments, a normalized value is obtained, thereby eliminating the differences in dimensions and numerical ranges. Finally, the data is stored in the format of "pipe segment number - geometric parameter type - normalized value - original value - unit" to form physical geometric attributes that can be directly input into the subsequent feature extraction module.

[0030] The beneficial effects are as follows: Employing multi-channel data collection methods, combined with industry standards to avoid parameter omissions, ensures the integrity and compliance of raw data, providing an accurate foundation for subsequent analysis, reducing the risk of optimization failure due to data issues, transforming unstructured layout information into structured topology data, clarifying duct geometric relationships and quantifying parameters, providing support for fluid dynamics constraint analysis, reducing calculation bias, transforming fuzzy performance requirements into standardized constraints, clarifying parameter optimization boundaries, ensuring that the manufactured ducts meet usage requirements, avoiding functional deficiencies, achieving a precise correlation between layout topology and performance requirements, ensuring that initial geometric attributes meet basic design and fluid constraints, reducing the scope of subsequent optimization adjustments, improving efficiency, and eliminating differences in parameter units and ranges through dimensional unification and numerical normalization, avoiding subsequent calculation errors, and ensuring the consistency and accuracy of data processing.

[0031] The feature extraction module 102 is used to filter the physical geometric attributes and perform structured encoding on the denoised physical geometric attributes to obtain the multimodal feature vector of the physical geometric attributes. In this embodiment of the invention, when the feature extraction module performs filtering on the physical geometric attributes and structured encoding on the denoised physical geometric attributes to obtain the multimodal feature vector of the physical geometric attributes, it is specifically used for: The physical geometric attributes are subjected to time-frequency domain filtering to obtain the duct geometric feature sequence of the physical geometric attributes; Multi-scale feature fusion is performed on the geometric feature sequence of the duct to obtain a multi-scale feature representation of the prefabricated duct. The multi-scale feature representation is encoded using a graph neural network to obtain the spatial topological feature tensor of the prefabricated duct. Extract multiple topological feature vectors from the spatial topological feature tensor, and combine the multiple topological feature vectors into a topological feature vector group; The topological feature vector group is embedded in a high-dimensional manifold to obtain the multimodal feature vector of the physical geometric attribute.

[0032] First, various physical geometric attribute data are converted into time-domain signal sequences, and then decomposed into frequency-domain signals of different frequencies using Fourier transform. Next, an adaptive threshold is set to remove high-frequency noise signals in the frequency domain and abnormal abrupt signals in the time domain. Finally, the filtered time-domain signal and the frequency-domain signal are reversed to reconstruct the duct geometric feature sequence with noise interference removed and data fluctuations stable. This sequence completely preserves the true data characteristics of the core geometric parameters of the duct. A top-down multi-scale feature fusion strategy is adopted: first, the macro-scale features are normalized to obtain a standardized overall feature vector; then, the micro-scale features are locally enhanced to highlight key details; weights are assigned to features of different scales according to the duct performance requirements; finally, the weighted macro-scale features and micro-scale features are spliced ​​together to form a multi-scale feature representation that covers both the whole and the part and takes into account both major and minor features. A graph convolutional neural network is used to encode multi-scale feature representations: First, a spatial topology graph of the duct is constructed, with each discrete segment, inflection point, and cross-section of the duct as a node in the graph, and the node attributes as the corresponding multi-scale feature vectors; the spatial connections between nodes are used as edges in the graph, and the weights of the edges are determined by the physical distance and geometric similarity between nodes; then, through 2-3 layers of graph convolution operations, the features of each node are aggregated and calculated with the features of neighboring nodes to realize the transmission and fusion of spatial association information; finally, all the encoded node features are arranged in spatial order to form a spatial topology feature tensor with uniform dimensions and containing spatial topology association information. Based on the structural division rules of the duct, three independent types of topological feature vectors are separated from the spatial topological feature tensor: discrete segment feature vector, inflection point feature vector, and cross-sectional feature vector. Then, according to the spatial path order of the duct, the discrete segment feature vector, the inflection point feature vectors at both ends, and the cross-sectional feature vector corresponding to the same discrete segment are grouped to form a subset of local topological feature vectors with the discrete segment as the core. Finally, all local topological feature vector subsets are combined in an orderly manner according to the overall path of the duct to form a topological feature vector group that completely covers the spatial topology of the duct. The topological feature vector group is normalized to eliminate the dimensional differences between different types of feature vectors. Then, the similarity between each feature vector in the high-dimensional space is calculated to construct a feature similarity matrix, which maps the high-dimensional feature vectors to a low-dimensional feature space of 32-64 dimensions. During the mapping process, the original similarity relationship between the feature vectors is maintained to ensure that the core feature information is not lost. Finally, the mapped low-dimensional feature vectors are integrated according to the original sequence to obtain a multimodal feature vector that combines spatial topological features, geometric size features, and correlation features. This vector can be directly input into the subsequent voltage drop calculation module for processing.

[0033] The beneficial effects are as follows: joint time-frequency domain filtering eliminates invalid interference, ensuring the accuracy and stability of the duct geometric feature sequence, providing high-quality data for multi-scale fusion and topology coding, and avoiding feature distortion. Multi-scale feature fusion covers both overall structure and local detailed features, breaking the limitations of a single scale, providing rich and complete feature inputs for spatial topology coding, transforming spatial topological relationships into structured tensors, realizing the quantification of spatial correlation features, solving the problem that traditional methods struggle to capture geometric correlations, extracting and orderly combining feature vectors by type, preserving the independence of local topological features and the overall spatial logic, facilitating subsequent high-dimensional embedding processing, improving targeting, and achieving dimensionality reduction while preserving core correlations, integrating multiple types of features to form standardized multimodal vectors, reducing computational complexity, and adapting to the needs of subsequent modules.

[0034] The pressure drop calculation module 103 is used to perform forward propagation of the multimodal feature vector according to the consistency constraint in fluid mechanics to obtain the pressure drop gradient distribution of the prefabricated air duct. In this embodiment of the invention, when the pressure drop calculation module performs forward propagation of the multimodal feature vector based on consistency constraints in fluid mechanics to obtain the pressure drop gradient distribution of the prefabricated duct, it is specifically used for: The multimodal feature vectors are spatially discretized to obtain the discretized spatial data of the prefabricated air duct. The flow field characteristics are evolved from the discretized spatial data to obtain the internal velocity field distribution of the prefabricated duct.

[0035] The pressure drop calculation module is also used for: Based on the preset wall normal gradient distribution, the field characteristics of the internal velocity field distribution are analyzed to obtain the velocity gradient distribution of the prefabricated air duct. The formula for calculating the pressure drop gradient distribution is as follows: ; In the formula, The pressure drop gradient distribution is... The total number of discrete segments in the spatial topological feature tensor. The first one obtained based on the physical geometric properties The cross-sectional area of ​​each discrete segment The first one defined based on the path topology data The spatial area occupied by each discrete section of the air duct. For boundary operators, denote the spatial region after the boundary operator. Take the boundary. For the first The wall boundaries of each discrete segment are determined based on the path topology data. The wall shear stress is obtained by converting the viscous stress relationship based on the velocity gradient distribution. Let the area of ​​the infinitesimal element on the wall boundary be . The first one obtained based on the path topology data The path element vector of each discrete segment Indicates the first The surface integral of the shear stress on a discrete segment of the wall. This represents the cumulative summation along the path for all discrete segments based on the path topology data; The velocity gradient distribution is transformed into a viscous stress relationship to obtain the shear stress distribution of the velocity gradient distribution; The pressure drop gradient distribution of the prefabricated duct is obtained by performing friction resistance accumulation processing on the shear stress distribution.

[0036] The core function of the pressure drop calculation module is to perform forward propagation calculation on the multimodal feature vector that integrates the physical geometric properties and path topology data of the prefabricated air duct, based on the consistency constraint rules of mass conservation, momentum conservation, and energy conservation in fluid mechanics, and finally output the pressure drop gradient distribution of the prefabricated air duct. The pressure drop calculation module first performs spatial discretization on the input multimodal feature vector. Specifically, the finite volume method is used to divide the continuous three-dimensional space of the prefabricated duct corresponding to the multimodal feature vector into... For each discrete segment, which consists of non-overlapping, clearly defined discrete units, its cross-sectional area, determined based on physical geometric properties, is extracted. At the same time, it clarifies the occupied space area of ​​each discrete segment based on the path topology data. This ultimately results in discretized spatial data containing the geometric parameters, spatial locations, and characteristic attributes of all discrete segments. Based on the discretized spatial data obtained above, the pressure drop calculation module further performs flow field characteristic evolution. The specific process is as follows: the geometric parameters and fluid property parameters of each discrete segment in the discretized spatial data are substituted into the Navier-Stokes equations to simulate and calculate the fluid flow state in each discrete unit. The parameters such as flow velocity and pressure are iteratively optimized step by step until the convergence condition is met, and finally the flow velocity magnitude and direction distribution data of each discrete unit inside the prefabricated duct are obtained, that is, the internal velocity field distribution. The pressure drop calculation module calls the preset wall normal gradient distribution data to perform field characteristic analysis on the above internal velocity field distribution; specific operation: calculate the rate of change of velocity in each discrete unit relative to the wall normal, combine the velocity difference between adjacent discrete units to determine the velocity variation law along the spatial position in each discrete segment, and finally obtain the velocity gradient distribution of the prefabricated air duct. The pressure drop gradient distribution of prefabricated ductwork is calculated using the following formula: ; The final pressure drop gradient distribution to be determined reflects the gradient change of pressure along the flow path within the duct. This represents the total number of discrete segments in the spatial topological feature tensor, which is consistent with the number of discrete segments obtained by spatial discretization. For the first The cross-sectional area of ​​each discrete segment comes from the spatial discretization process. For the first The spatial area occupied by each discrete segment of the duct is defined by the path topology data; It is a boundary operator that acts on a spatial region. Its wall boundary was then obtained. ; dS represents the wall shear stress, derived from the subsequent viscous stress relationship; dS represents the wall boundary. The area of ​​the infinitesimal element on it; Indicates the first The wall boundary of each discrete segment Wall shear stress of all infinitesimal elements The total effect value of the wall shear stress of the discrete segment is obtained by performing surface integral calculation; For the first The path element vectors of each discrete segment are determined by the path topology data; This indicates that the calculation results of all discrete segments are cumulatively summed along the path according to the discrete segment order determined by the path topology data, to obtain the overall pressure drop gradient distribution.

[0037] The pressure drop calculation module performs a viscous stress relationship conversion on the velocity gradient distribution obtained above. The specific process is as follows: the velocity change rate data of each discrete unit in the velocity gradient distribution is multiplied with the known fluid dynamic viscosity to obtain the magnitude, direction and distribution location of the shear stress corresponding to each discrete unit, i.e., the shear stress distribution. The pressure drop calculation module performs friction loss accumulation processing on the shear stress distribution. Specifically, it extracts the shear stress distribution data for each discrete segment sequentially, according to the order of the discrete segments determined by the path topology data, and combines this data with the cross-sectional area of ​​that discrete segment. Path infinitesimal vector The contribution of the resistance generated by shear stress in the discrete segment to the pressure drop is calculated, and then the contribution values ​​of all discrete segments are accumulated and summed along the flow path to finally obtain the overall pressure drop gradient distribution of the prefabricated duct.

[0038] The beneficial effects are as follows: Using fluid dynamics consistency constraints as the calculation basis ensures that the pressure drop gradient distribution results conform to fundamental laws, fundamentally avoiding deviations caused by unreasonable basis. Spatial discretization divides the continuous duct space into numerically calculable discrete units, clarifying the geometric parameters and spatial regions of each unit, providing accurate data support for flow field evolution. Flow field characteristic evolution is based on the Navier-Stokes equations to simulate real flow, accurately obtaining the internal velocity field distribution, freeing pressure drop calculations from empirical formulas and improving reliability. Combined with preset wall normal gradient analysis of the velocity field, the velocity change law is accurately discovered, and the obtained velocity gradient distribution provides a direct basis for viscous stress conversion, reducing errors. The pressure drop calculation formula is clarified and key parameters are incorporated, comprehensively considering the influence of geometry and flow field on pressure drop, avoiding calculation ambiguity, and ensuring that the results reflect the actual pressure drop situation. Based on Newton's law of viscosity, the conversion of velocity gradient to shear stress is completed, establishing the correlation between flow field and resistance, providing accurate physical quantities for friction loss accumulation. Friction loss accumulation is superimposed according to the path sequence of each discrete segment's contribution, comprehensively considering path influence factors, accurately reflecting the pressure drop change law, and providing a technical basis for duct structure optimization.

[0039] The parameter optimization module 104 is used to iteratively adjust the geometric performance characteristic parameters in the multimodal feature vector based on the backpropagation optimization of the pressure drop gradient distribution to obtain the target manufacturing guidance parameters of the prefabricated air duct. In this embodiment of the invention, when the parameter optimization module performs backpropagation optimization based on the pressure drop gradient distribution and iteratively adjusts the geometric performance characteristic parameters in the multimodal feature vector to obtain the target manufacturing guidance parameters of the prefabricated duct, it is specifically used for: By performing an adjoint differential sensitivity analysis on the pressure drop gradient distribution, the adjoint gradient field of the pressure drop gradient distribution is obtained. Based on the accompanying gradient field, the geometric performance characteristic parameters are optimized by backpropagation to obtain the geometric parameter correction amount of the prefabricated air duct. The formula for calculating the geometric parameter correction is as follows: ; = ; In the formula, and The first and The geometric performance characteristic parameters of the next iteration. The preset learning rate, For the first Sensitivity coefficient at the next iteration This is the cumulative momentum of the sensitivity coefficient. The adaptive adjustment amount of the sensitivity coefficient is denoted as , where and To estimate the attenuation rate using moments, It is the numerical stability constant; The geometric parameter correction amount is optimized by adaptive moment estimation to obtain the parameter update value of the geometric parameter correction amount; By projecting the updated parameter values ​​onto the manufacturing process constraints, feasible parameter solutions for the updated parameter values ​​are obtained.

[0040] The parameter optimization module is also used for: The feasible parameters are demapped to the convergence solution set defined by the convergence criterion to obtain the optimized parameters of the prefabricated air duct. The optimization parameters are subjected to Pareto screening to obtain a set of candidate parameters for the optimization parameters; A manufacturing feasibility assessment is performed on the candidate parameter set to obtain the target manufacturing guidance parameters for the prefabricated air duct.

[0041] Based on the pressure drop gradient distribution of the prefabricated duct output by the pressure drop calculation module, the geometric performance characteristics parameters of the duct contained in the multimodal feature vector are iteratively adjusted, and the final output is the target manufacturing guidance parameters of the prefabricated duct that can directly guide the production. A percutaneous differential sensitivity analysis is performed on the previously obtained pressure drop gradient distribution. The specific steps are as follows: Using the percutaneous equation method, a correlation model between the pressure drop gradient and each geometric performance parameter is constructed. The influence of small changes in each geometric parameter on the pressure drop gradient distribution is calculated by solving the percutaneous equation. Simultaneously, the sensitivity information of all parameters is integrated according to their spatial distribution to form a percutaneous gradient field reflecting the strength of the "parameter-pressure drop" correlation. Based on the percutaneous gradient field obtained above, the parameter optimization module performs backpropagation optimization on the geometric performance parameters. The core process is to calculate the geometric parameter correction amount using a preset formula. The specific steps are as follows: : No. The current geometric performance characteristics parameters at the next iteration; Preset learning rate; , : Moment estimation of attenuation rate; Numerical stability constant; : No. Sensitivity coefficient at the next iteration; No. The momentum accumulation of the sensitivity coefficient at the next iteration is calculated according to the formula... = Calculate, where, This is the accumulated momentum from the previous iteration, initially set to 0, used to smooth the direction of parameter adjustments and reduce oscillations; : No. The adaptive adjustment of the sensitivity coefficient in the next iteration is calculated according to the formula. Calculate, where, This is the adaptive adjustment amount from the previous iteration, with an initial value of 0. It is used to dynamically adjust the effective step size of the learning rate to adapt to the sensitivity differences of different parameters. Substitute the above parameters and apply the formula. Calculate the first Geometric parameter corrections for each iteration; The calculated geometric parameter corrections are optimized using adaptive moment estimation. Specifically, this involves optimizing the momentum accumulation based on the accumulated momentum from each iteration. and adaptive adjustment amount Dynamically adjust the update step size of the correction amount—when the parameter sensitivity is high, by... Increasing the effective step size accelerates parameter tuning; when the parameters are close to the optimal solution, by... Smoothly adjust direction, reduce step size to avoid overshoot; simultaneously... , The weight allocation balances "historical adjustment experience" and "current sensitivity information" to ultimately obtain parameter update values ​​for geometric parameter corrections that are more stable and converge faster. The preset manufacturing process constraints are invoked, and the parameter update values ​​are projected onto the manufacturing process constraints. Specifically, the parameter update values ​​are mapped to the feasible domain that is jointly defined by all constraints. If the parameter update values ​​exceed the constraint range, they are corrected according to the boundary values ​​of the feasible domain. Finally, a feasible parameter solution that satisfies all manufacturing process constraints is obtained. A convergence criterion is preset, mapping the feasible parameter solutions to the convergent solution set defined by the criterion. Specifically, each feasible parameter solution is checked to see if it satisfies the convergence criterion. If it does, it is added to the convergent solution set; otherwise, the process returns to the "Adjoint Differential Sensitivity Analysis" step for the next iteration until optimized parameters that meet the convergence criterion are obtained. The optimized parameters obtained above are subjected to Pareto screening. The specific process is as follows: For the multi-objective requirements of duct design, the non-dominated ranking method is used to rank the optimized parameters—if no other parameter is better than the current parameter in all objectives, then that parameter is the Pareto optimal solution; all Pareto optimal solutions are screened to form a candidate parameter set for the optimized parameters; A manufacturing feasibility assessment is conducted on the candidate parameter set. Specifically, each candidate parameter is quantitatively scored based on four dimensions: processing difficulty, mass production adaptability, quality stability, and cost controllability. Parameters with scores below a preset threshold are eliminated, and the parameters with the best overall feasibility are retained, ultimately yielding the target manufacturing guidance parameters for the prefabricated duct.

[0042] The beneficial effects are as follows: Backpropagation optimization directly links parameter adjustment with voltage drop performance, ensuring that the optimization direction aligns with engineering realities. Simultaneously, with "manufacturing-oriented parameters" as the goal, it connects with the production process, enhancing engineering value. Accompanied by differential sensitivity analysis, it accurately captures the degree and direction of parameter influence on voltage drop, forming an accompanying gradient field to avoid blind adjustments, reduce iterations, and improve optimization accuracy and efficiency. The geometric parameter correction formula incorporates key factors such as learning rate and momentum accumulation, smoothing the adjustment direction and reducing oscillations while dynamically adapting to sensitivity differences to accelerate convergence, ensuring iteration stability and repeatability. Adaptive moment estimation optimization dynamically adjusts the parameter update step size, solving the problem of "overshooting and slow convergence" with fixed step sizes, enabling parameters to quickly approximate the target voltage. Near-optimal solution with stable convergence enhances robustness. Manufacturing process constraint projection maps parameter update values ​​to the feasible region, eliminating over-constraint parameters, preventing design-production disconnect, reducing rework costs, and ensuring parameter manufacturability. Convergence screening judges whether parameters meet preset criteria; if not, iterative optimization ensures the stability and reliability of the final optimized parameters, avoiding result failure due to non-convergence. Pareto screening selects the optimal solution for multiple objective requirements, forming a candidate parameter set to balance requirements such as "low pressure drop, small volume, and low cost," avoiding performance imbalance caused by single-objective optimization. Manufacturing feasibility assessment quantifies and scores from multiple dimensions, eliminating low-fit parameters and retaining the comprehensive optimal solution, achieving a balance between "high performance and ease of manufacturing," and enhancing industrialization value.

[0043] The process mapping module 105 is used to map the target manufacturing guidance parameters to the material-process mapping knowledge base to obtain the CNC machining instruction set of the prefabricated air duct; In this embodiment of the invention, when the process mapping module executes the mapping of the target manufacturing guidance parameters to the material-process mapping knowledge base to obtain the CNC machining instruction set of the prefabricated air duct, it is specifically used for: The target manufacturing guidance parameters are decomposed to obtain the parameter feature components of the prefabricated air duct; Based on a pre-set material-process knowledge base, multi-dimensional constraint matching is performed on the parameter feature components to obtain an initial process scheme for the parameter feature components. The processing paths and parameters in the initial process scheme are optimized collaboratively to obtain the optimized process configuration of the prefabricated air duct; Numerical control instructions are generated for the optimized process configuration to obtain the numerical control machining instruction set for the prefabricated air duct.

[0044] The core function of the process mapping module is to receive the target manufacturing guidance parameters of the prefabricated duct output from the preceding process and map them to the preset material-process mapping knowledge base. Through the complete process of "feature decomposition → multi-dimensional constraint matching → collaborative optimization → CNC instruction generation", the module finally outputs a set of machining instructions that can directly drive CNC equipment. Based on four dimensions—geometric shape, material properties, processing accuracy, and performance requirements—the target manufacturing guidance parameters are decomposed into quantifiable parameter feature components: the geometric shape is clearly defined as a pipe diameter of 300mm, a bend of 60°, and a total length of 5000mm; the material properties are 304 stainless steel, suitable for cold bending processing; the accuracy requirements are a dimensional tolerance of ±0.05mm and a wall thickness reduction rate of ≤5%; and the performance requirements are an inner wall roughness Ra≤1.6μm, providing refined input for subsequent precise process matching.

[0045] Based on the material-process mapping knowledge base, the process mapping module performs multi-dimensional constraint matching on the parameter feature components: the material dimension matches "cold bending + CNC turning + fine polishing"; the geometric dimension matches "WC67Y-100 CNC pipe bending machine + φ300mm die"; the accuracy dimension matches "cutting speed 80m / min, feed rate 0.2mm / r"; and the performance dimension matches "2 fine turning passes + 1 polishing pass". Finally, the initial process scheme is integrated to meet the basic requirements of the target parameters. The processing path and parameters of the initial process plan were optimized collaboratively: the bending trajectory was optimized into a continuous trajectory of "starting point → 2000mm straight line → 60° arc → 2500mm straight line → ending point"; the cold bending speed was adjusted to 3mm / s and the fine turning feed was finely adjusted to 0.15mm / r to meet the trajectory optimization requirements; a deburring process was added, resulting in an optimized process configuration: the equipment and tooling consisted of a WC67Y-100 pipe bending machine + a special mold + a deburring tooling, and the process flow was "cold bending → turning → deburring → fine turning → polishing"; Optimize process configuration to generate CNC instructions: Extract key data such as machining path coordinates, machining parameters, and tool selection, convert them into standardized instructions according to G / M code specifications, and then integrate and sort them according to the process sequence to form a complete and logically coherent CNC machining instruction set, which can be directly input into the WC67Y-100 CNC pipe bending machine for processing.

[0046] The beneficial effects include establishing a direct link between design parameters and CNC machining, transforming optimized target manufacturing-oriented parameters into executable instructions, avoiding the disconnect between design and production, and improving the ability to implement industrialization. Target parameter feature decomposition breaks down comprehensive parameters into independent components, solving the difficulty of process matching caused by mixed dimensions, enabling accurate identification of requirements in each dimension, and improving matching accuracy. Multi-dimensional constraint matching combines conditions such as materials, geometry, precision, and performance, relying on a material-process knowledge base to screen initial solutions, avoiding the irrationality of single-dimensional matching, reducing trial and error costs. Machining paths and parameters are co-optimized, optimizing trajectories to reduce redundancy and adapting parameters to avoid conflicts. At the same time, necessary processes are added, improving machining efficiency and quality, reducing equipment wear. CNC instruction generation converts optimized processes into standardized code according to specifications, which can directly drive equipment, reducing manual programming errors, ensuring machining consistency, and adapting to mainstream equipment, enhancing versatility.

[0047] The adaptive optimization module 106 is used to execute the CNC machining instruction set, obtain real-time data during the duct processing, and optimize and update the material-process mapping knowledge base based on the real-time data.

[0048] In this embodiment of the invention, when the adaptive optimization module executes the CNC machining instruction set to obtain real-time data during the duct machining process, and optimizes and updates the material-process mapping knowledge base based on the real-time data, it is specifically used for: The CNC machining instruction set is digitally processed to obtain real-time processing status data of the prefabricated air duct; Multi-source data fusion analysis is performed on the real-time processing status data to obtain the quality characteristic parameters of the prefabricated air duct during the processing. Key quality indicators are extracted from the quality characteristic parameters and their matching degree is evaluated with the material-process mapping knowledge base to obtain the optimization requirement indicators of the material-process mapping knowledge base. The knowledge base is incrementally updated based on the optimization requirement indicators to obtain the optimized material-process mapping knowledge base for the prefabricated air duct.

[0049] The core function of the adaptive optimization module is to receive the CNC machining instruction set output by the process mapping module, drive the CNC machining equipment to execute the instruction set to complete the prefabricated air duct processing, synchronously collect real-time data during the processing, and then dynamically optimize and update the material-process mapping knowledge base based on the real-time data, forming a closed-loop optimization mechanism of "processing-feedback-iteration".

[0050] The CNC machining instruction set is transmitted to equipment such as the WC67Y-100 CNC pipe bending machine and CNC turning machine, driving the equipment to execute digital machining according to the instructions: following the process flow of "cold bending → turning → deburring → precision turning → polishing", the air duct with a diameter of 300mm and a bend of 60° is processed; at the same time, through the displacement sensor, pressure sensor, roughness tester and other equipment, real-time processing status data during the processing is collected, including processing parameter data, equipment operation data, quality inspection data and so on.

[0051] The collected multi-source real-time processing status data is fused and analyzed: first, the data is denoised, then the processing parameter data is correlated and matched with the quality inspection data and equipment operation data, and finally the quality feature parameters including wall thickness reduction rate, inner wall roughness, dimensional tolerance, bending angle deviation, etc. are extracted to comprehensively reflect the quality status of the processing process.

[0052] Key quality indicators are extracted from the quality characteristic parameters. The matching degree of these indicators is evaluated with the preset quality standards corresponding to the combination of "304 stainless steel - cold bending + CNC turning" in the material-process mapping knowledge base. If the actual wall thickness reduction rate is 6.2% and the inner wall roughness Ra=1.8μm, the quantitative matching degree is 85%. The optimization requirement indicators are determined to be "adjusting the cold bending speed and the fine turning feed to make the wall thickness reduction rate ≤5% and the inner wall roughness Ra≤1.6μm".

[0053] To address this optimization requirement, the material-process mapping knowledge base was incrementally updated: only the parameter entry corresponding to "304 stainless steel - 300mm pipe diameter - cold bending" was corrected, adjusting the original cold bending speed from 3mm / s to 2.8mm / s and the original finishing feed rate from 0.15mm / r to 0.12mm / r. Quality standard verification data corresponding to "cold bending speed 2.8mm / s + finishing feed rate 0.12mm / r" was also added. All other unaffected entries remained unchanged, resulting in an optimized material-process mapping knowledge base, providing a more accurate basis for process matching in subsequent similar duct processing.

[0054] The beneficial effects are as follows: A closed loop of "executing instructions – collecting data – updating the knowledge base" is constructed, solving the problem of decreased adaptability in traditional static knowledge bases. This ensures the knowledge base continuously aligns with actual processing scenarios, improving the accuracy of subsequent process matching. While driving equipment to execute instructions, real-time processing, equipment, and quality data are collected through multiple types of sensors, providing a true and objective basis for knowledge base optimization and avoiding the blindness of theoretical deduction. Multi-source data fusion analysis first removes noise and then correlates and matches data, uncovering the intrinsic relationship between parameters, equipment, and quality to form comprehensive and accurate quality characteristic parameters, laying the foundation for subsequent evaluation. Key quality indicators are extracted and the knowledge base matching degree is evaluated, quantifying the deviation between reality and standards, clarifying the direction of knowledge base optimization, avoiding blind updates, and improving optimization efficiency. Incremental updates only correct knowledge base entries related to requirements, without overall reconstruction, ensuring that the knowledge base adapts to actual needs while avoiding compatibility issues and redundant operations, thus improving iteration efficiency and stability.

[0055] Reference Figure 2 The diagram shown is a flowchart illustrating an intelligent processing method for prefabricated ductwork based on digital modeling, according to an embodiment of the present invention. In this embodiment, the intelligent processing method for prefabricated ductwork based on digital modeling includes: S1: Used to obtain the layout parameters and performance requirements of prefabricated air ducts, and to obtain the physical geometric properties of prefabricated air ducts; S2: Used to filter the physical geometric attributes and perform structured encoding on the denoised physical geometric attributes to obtain the multimodal feature vector of the physical geometric attributes; S3: Used to perform forward propagation of the multimodal feature vector according to the consistency constraints in fluid mechanics to obtain the pressure drop gradient distribution of the prefabricated air duct; S4: Used for backpropagation optimization based on the pressure drop gradient distribution to iteratively adjust the geometric performance characteristic parameters in the multimodal feature vector to obtain the target manufacturing guidance parameters of the prefabricated duct. S5: Used to map the target manufacturing guidance parameters to the material-process mapping knowledge base to obtain the CNC machining instruction set of the prefabricated air duct; S6: Used to execute the CNC machining instruction set, obtain real-time data during the duct processing, and optimize and update the material-process mapping knowledge base based on the real-time data.

[0056] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0057] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence is the theory, method, technology, and application system that uses digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0058] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A prefabricated air duct intelligent processing system based on digital modeling, characterized in that, The system includes a parameter acquisition module, a feature extraction module, a voltage drop calculation module, a parameter optimization module, a process mapping module, and an adaptive optimization module, wherein: The parameter acquisition module is used to acquire the layout parameters and performance requirements of the prefabricated air duct, and obtain the physical geometric properties of the prefabricated air duct. The feature extraction module is used to filter the physical geometric attributes and perform structured encoding on the denoised physical geometric attributes to obtain the multimodal feature vector of the physical geometric attributes. The pressure drop calculation module is used to perform forward propagation of the multimodal feature vector according to the consistency constraints in fluid mechanics to obtain the pressure drop gradient distribution of the prefabricated air duct. The parameter optimization module is used to iteratively adjust the geometric performance characteristic parameters in the multimodal feature vector based on backpropagation optimization of the pressure drop gradient distribution, to obtain the target manufacturing guidance parameters of the prefabricated duct, specifically for: By performing an adjoint differential sensitivity analysis on the pressure drop gradient distribution, the adjoint gradient field of the pressure drop gradient distribution is obtained. Based on the accompanying gradient field, the geometric performance characteristic parameters are optimized by backpropagation to obtain the geometric parameter correction amount of the prefabricated air duct. The formula for calculating the geometric parameter correction is as follows: ; = , ; In the formula, and The geometric performance characteristic parameters are respectively for the t-th and t+1-th iterations. The preset learning rate, Let be the sensitivity coefficient at the t-th iteration. This is the cumulative momentum of the sensitivity coefficient. The adaptive adjustment amount of the sensitivity coefficient is denoted as , where and To estimate the attenuation rate using moments, Here is the numerical stability constant. This is the accumulated momentum from the previous iteration. This is the adaptive adjustment amount from the previous iteration. and For the first The moment estimate of the decay rate in the next iteration; The geometric parameter correction amount is optimized by adaptive moment estimation to obtain the parameter update value of the geometric parameter correction amount; By projecting the updated parameter values ​​under manufacturing process constraints, a feasible parameter solution for the updated parameter values ​​is obtained. The process mapping module is used to map the target manufacturing guidance parameters to the material-process mapping knowledge base to obtain the CNC machining instruction set of the prefabricated air duct; The adaptive optimization module is used to execute the CNC machining instruction set, obtain real-time data during the duct processing, and optimize and update the material-process mapping knowledge base based on the real-time data.

2. The intelligent processing system for prefabricated air ducts based on digital modeling as described in claim 1, characterized in that, When the parameter acquisition module obtains the layout parameters and performance requirements of the prefabricated ductwork and acquires the physical geometric properties of the prefabricated ductwork, it is specifically used for: Obtain the layout parameters and performance requirements of prefabricated air ducts; The layout parameters of the prefabricated air duct are digitally analyzed to obtain the path topology data of the prefabricated air duct. The performance requirements are extracted in a structured manner to obtain the fluid dynamic constraints of the prefabricated air duct; Based on the fluid dynamics constraints, a geometric correlation analysis is performed on the path topology data to obtain the initial geometric properties of the prefabricated duct. The initial geometric properties are normalized to obtain the physical geometric properties of the prefabricated duct.

3. The intelligent processing system for prefabricated air ducts based on digital modeling as described in claim 1, characterized in that, When the feature extraction module performs filtering on the physical geometric attributes and structured encoding on the denoised physical geometric attributes to obtain the multimodal feature vector of the physical geometric attributes, it is specifically used for: The physical geometric attributes are subjected to time-frequency domain filtering to obtain the duct geometric feature sequence of the physical geometric attributes; Multi-scale feature fusion is performed on the geometric feature sequence of the duct to obtain a multi-scale feature representation of the prefabricated duct. The multi-scale feature representation is encoded using a graph neural network to obtain the spatial topological feature tensor of the prefabricated duct. Extract multiple topological feature vectors from the spatial topological feature tensor, and combine the multiple topological feature vectors into a topological feature vector group; The topological feature vector group is embedded in a high-dimensional manifold to obtain the multimodal feature vector of the physical geometric attribute.

4. The intelligent processing system for prefabricated air ducts based on digital modeling as described in claim 2, characterized in that, When the pressure drop calculation module performs forward propagation of the multimodal feature vector based on consistency constraints in fluid mechanics to obtain the pressure drop gradient distribution of the prefabricated duct, it is specifically used for: The multimodal feature vectors are spatially discretized to obtain the discretized spatial data of the prefabricated air duct. The flow field characteristics are evolved from the discretized spatial data to obtain the internal velocity field distribution of the prefabricated duct.

5. The intelligent processing system for prefabricated air ducts based on digital modeling as described in claim 4, characterized in that, The pressure drop calculation module is also used for: Based on the preset wall normal gradient distribution, the field characteristics of the internal velocity field distribution are analyzed to obtain the velocity gradient distribution of the prefabricated air duct. The formula for calculating the pressure drop gradient distribution is as follows: ; In the formula, The pressure drop gradient distribution is... Let be the total number of discrete segments in the spatial topological feature tensor. The first one obtained based on the physical geometric properties The cross-sectional area of ​​each discrete segment The first one defined based on the path topology data The spatial area occupied by each discrete section of the air duct. For boundary operators, denote the spatial region after the boundary operator. Take the boundary. For the first The wall boundaries of each discrete segment are determined based on the path topology data. The wall shear stress is obtained by converting the viscous stress relationship based on the velocity gradient distribution. Let the area of ​​the infinitesimal element on the wall boundary be . The first one obtained based on the path topology data The path element vector of each discrete segment Indicates the first The surface integral of the shear stress on a discrete segment of the wall. This represents the cumulative summation along the path for all discrete segments based on the path topology data; The velocity gradient distribution is transformed into a viscous stress relationship to obtain the shear stress distribution of the velocity gradient distribution; The pressure drop gradient distribution of the prefabricated duct is obtained by performing friction resistance accumulation processing on the shear stress distribution.

6. The intelligent processing system for prefabricated air ducts based on digital modeling as described in claim 1, characterized in that, The parameter optimization module is also used for: The feasible parameters are demapped to the convergence solution set defined by the convergence criterion to obtain the optimized parameters of the prefabricated air duct. The optimization parameters are subjected to Pareto screening to obtain a set of candidate parameters for the optimization parameters; A manufacturing feasibility assessment is performed on the candidate parameter set to obtain the target manufacturing guidance parameters for the prefabricated air duct.

7. The intelligent processing system for prefabricated air ducts based on digital modeling as described in claim 1, characterized in that, When the process mapping module maps the target manufacturing guidance parameters to the material-process mapping knowledge base to obtain the CNC machining instruction set for the prefabricated air duct, it is specifically used for: The target manufacturing guidance parameters are decomposed to obtain the parameter feature components of the prefabricated air duct; Based on a pre-set material-process knowledge base, multi-dimensional constraint matching is performed on the parameter feature components to obtain an initial process scheme for the parameter feature components. The processing paths and parameters in the initial process scheme are optimized collaboratively to obtain the optimized process configuration of the prefabricated air duct; Numerical control instructions are generated for the optimized process configuration to obtain the numerical control machining instruction set for the prefabricated air duct.

8. The intelligent processing system for prefabricated air ducts based on digital modeling as described in claim 1, characterized in that, When the adaptive optimization module executes the CNC machining instruction set to obtain real-time data during the duct machining process, and optimizes and updates the material-process mapping knowledge base based on the real-time data, it is specifically used for: The CNC machining instruction set is digitally processed to obtain real-time processing status data of the prefabricated air duct; Multi-source data fusion analysis is performed on the real-time processing status data to obtain the quality characteristic parameters of the prefabricated air duct during the processing. Key quality indicators are extracted from the quality characteristic parameters and their matching degree is evaluated with the material-process mapping knowledge base to obtain the optimization requirement indicators of the material-process mapping knowledge base. The knowledge base is incrementally updated based on the optimization requirement indicators to obtain the optimized material-process mapping knowledge base for the prefabricated air duct.

9. A method for intelligent processing of prefabricated air ducts based on digital modeling, characterized in that: The method for using the intelligent processing system for prefabricated ductwork based on digital modeling as described in claim 1 includes: S1: Obtain the layout parameters and performance requirements of the prefabricated air duct, and obtain the physical geometric properties of the prefabricated air duct; S2: Filter the physical geometric attributes and perform structured encoding on the denoised physical geometric attributes to obtain the multimodal feature vector of the physical geometric attributes; S3: Based on the consistency constraints in fluid mechanics, the multimodal feature vector is forward propagated to obtain the pressure drop gradient distribution of the prefabricated duct. S4: Based on the backpropagation optimization of the pressure drop gradient distribution, the geometric performance characteristic parameters in the multimodal feature vector are iteratively adjusted to obtain the target manufacturing guidance parameters of the prefabricated air duct; S5: Map the target manufacturing guidance parameters to the material-process mapping knowledge base to obtain the CNC machining instruction set for the prefabricated air duct; S6: Execute the CNC machining instruction set to obtain real-time data during the duct processing, and optimize and update the material-process mapping knowledge base based on the real-time data.

Images