Multi-mode intelligent cost accounting method and system for casting parts

Abstract

The invention discloses a casting part multi-mode intelligent cost accounting method and system, and relates to the technical field of casting engineering and intelligent cost accounting. The casting part multi-mode intelligent cost accounting method comprises the steps that process information in casting engineering is extracted and processed in a structured mode, and missing information of a plane drawing is automatically supplemented; matching a target casting process route and key parameters based on the process information subjected to the structured processing; and calculating the basic total cost according to the target casting process route, and correcting the basic total cost in combination with the historical data deviation, the real-time market data and the order variables to obtain the corrected total cost. Unified automatic analysis of the plane drawing and the three-dimensional drawing is achieved, the problem that information of the plane drawing is incomplete is solved through image recognition and rule completion, and the limitation that only a single drawing type is supported is broken through; the casting process constraint extracted from the three-dimensional drawing is directly used as a core parameter of process matching, and the plane drawing indirectly maps process requirements through feature recognition, so that quantitative association of drawing information-process parameter-cost is realized.

Description

Technical Field

[0001] This invention relates to the fields of casting engineering and intelligent cost accounting technology, and more specifically to a multimodal intelligent cost accounting method and system for castings. Background Technology

[0002] With the intensification of global competition in the manufacturing industry, foundry companies need to respond quickly to customer quotation requests. After customers provide design drawings for castings, companies need to provide quotations that balance "cost accuracy" and "market competitiveness" within a short period of time. However, traditional methods of pricing castings have insurmountable flaws, and existing technologies cannot meet the specific needs of the foundry industry. Poor drawing compatibility: It can only process 3D models or rely on manual interpretation of 2D drawings (such as CAD 2D drawings and PDF drawings), and cannot achieve automated and unified processing of 2D / 3D drawings. Dimensions and process annotations of 2D drawings need to be extracted manually, which is inefficient. Insufficient information utilization: PMI information (such as tolerances and process constraints) in 3D drawings is not parsed at the system level, and annotation information (such as dimensional tolerances and material codes) in 2D drawings relies on manual input, which easily leads to the omission of key parameters (such as casting fillets and wall thickness limits). Subjectivity in process matching: Relying on engineers' experience to select casting processes (such as sand casting / die casting) without establishing a quantitative correlation between "drawing information - process parameters - cost" leads to a disconnect between process and cost accounting; Static cost accounting: It does not integrate dynamic data such as real-time material prices, regional production capacity, and historical process deviations, resulting in discrepancies between the quoted prices and the actual market situation, and it lacks the ability to adaptively adjust variables such as batch size and lead time.

[0003] In summary, existing technologies cannot solve the core problems of "low efficiency, large error, and poor process adaptability" in casting pricing, and there is an urgent need for an automatic pricing solution specifically designed for casting design drawings. Summary of the Invention

[0004] In view of this, the present invention provides a multimodal intelligent cost accounting method and system for castings to solve the problems in the background art.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: On one hand, this invention discloses a multimodal intelligent cost accounting method for castings, the specific steps of which include: S1. Receive 2D or 3D drawings, extract and structure the process information, and automatically supplement the missing information in the 2D drawings; S2. Based on the structured process information, match the target casting process route and key process parameters; S3. Calculate the basic total cost based on the target casting process route, including material cost, process cost and additional cost; S4. By combining historical data deviations, real-time market data, and order variables, the basic total cost is corrected to obtain the corrected total cost; S5. Generate a quotation result including cost details and process description based on the revised total cost.

[0006] Preferably, in the above-mentioned multimodal intelligent cost accounting method for castings, the specific steps of S1 are as follows: If the input is a 3D drawing, the PMI information is parsed and extracted, and converted into structured parameters, including geometric dimensions, tolerances, materials, and process constraints. If the input is a 2D drawing, the OCR system recognizes the dimensions, process symbols, and material codes, and the missing parameters are completed using a rule engine. Based on the default process requirements of the materials, a structured parameter set is generated.

[0007] Preferably, in the above-mentioned multimodal intelligent cost accounting method for castings, the structured parameters of S1 are input into the process knowledge graph, and the optimal casting process route and key parameters are matched through a machine learning model. The specific steps are as follows: S201, Construct or update the process knowledge graph based on the input structured process parameters, and clarify the nodes and relationships; S202, Utilize graph features for data embedding and representation learning, and train a machine learning model in combination with historical process data to learn the complex mapping between process routes and parameters; S203, For new input parameters, Retrieve and recommend the optimal process route and key parameter combination in the graph through model inference and similarity matching; S204, Iteratively optimize the model and knowledge graph using actual production feedback data.

[0008] Preferably, in the above-mentioned multimodal intelligent cost accounting method for castings, the casting cost is quantitatively calculated item by item to generate an initial cost estimate. The specific steps are as follows: The process-cost calculation model is invoked, and the total cost of the casting is calculated based on the S2 process route: material cost + mold cost + labor cost + auxiliary cost. The total material cost is calculated as follows: (mass of the casting body + mass of the gating and riser) × unit price of materials × (1 + material loss rate). Casting body mass = casting body volume × material density; gating and riser mass = gating and riser volume × material density. If the drawing does not have a gating and riser design, it is calculated by default as 10%-15% of the casting body mass and manually adjusted; material unit price: retrieved from the "real-time material price database"; material loss rate: preset according to the process. Different allocation methods are used depending on the mold type (disposable / reusable): One-time mold: Unit cost of mold = Cost of a single mold piece; Reusable molds: Unit cost of mold = (Mold manufacturing cost + Mold maintenance cost × Estimated service life) ÷ Estimated lifespan of mold; Labor cost calculation covers the entire casting process and is calculated by summing "labor hours of each process step × corresponding fee rate of each process step", as shown in the following formula: Total labor cost = Molding labor cost + Smelting labor cost + Casting labor cost + Cleaning labor cost + Heat treatment labor cost; Molding time = Number of castings required for molding ÷ Molding capacity Smelting time = (mass of casting body + mass of gating and riser) ÷ capacity of smelting equipment; Pouring time = Number of casting cavities × Pouring time per mold; Cleaning time = casting complexity coefficient × casting body mass ÷ cleaning equipment capacity; Heat treatment time = heat treatment furnace loading capacity ÷ casting body mass × holding time; Labor cost rate: Preset according to process steps.

[0009] Ancillary costs include energy consumption costs, testing costs, and other expenses, as detailed below: Energy cost = Σ (power of equipment in each process step × labor hours × electricity price); Testing cost = Unit price of testing type × Quantity of testing × Testing ratio; Other costs: including filter materials, release agents, and paint consumables, estimated at 1%-2% of the casting's mass.

[0010] Preferably, in the above-mentioned multimodal intelligent cost accounting method for castings, step S4 is as follows: S401 identifies and quantifies deviations in historical cost data, and calculates the average deviation rate of each cost item as a basic correction coefficient by comparing historical forecasts with actual values. S402 incorporates real-time market data and dynamically adjusts cost items that are strongly correlated with the market through a preset sensitivity model; S403 performs customized cost increase / decrease calculations based on specific variables of the current order; S404 combines the aforementioned historical deviation correction, market dynamic adjustment, and order variable impact, and applies them to the base total cost through an integrated calibration model, outputting the corrected total cost.

[0011] On the other hand, this invention discloses a multimodal intelligent cost accounting system for castings, which applies the above-mentioned method and includes: The multi-source drawing parsing module is used to receive 2D or 3D drawings, extract and structure the process-related information in them, including dimensional parameters, tolerance requirements, material information, process constraints and quality requirements. Missing information in 2D drawings is automatically supplemented based on a database of similar products. The intelligent process matching module has a built-in casting process knowledge graph. Based on the structured process-related information, it matches the target casting process route and key process parameters through a machine learning model. The large-scale process-cost calculation model includes a material cost sub-model, a process cost sub-model, and an additional cost sub-model, which is used to calculate the basic total cost based on the target casting process route and structured process information. The big data dynamic optimization module connects to the historical quotation database and real-time market data to perform deviation calibration, market adaptation, and order variable adjustment on the basic total cost, and outputs the corrected total cost. The quotation generation module generates a quotation result that includes cost details, process descriptions, and price suggestions based on the revised total cost.

[0012] Preferably, in the above-mentioned multimodal intelligent cost accounting system for castings, the intelligent process matching module has a built-in casting process knowledge graph, which includes the applicable conditions for 12 types of processes such as clay sand casting, resin sand casting, lost foam casting, investment casting, and die casting. Based on the structured parameters output by the multi-source drawing parsing module, the optimal process route is matched through fuzzy reasoning + machine learning model.

[0013] As can be seen from the above technical solutions, compared with the prior art, this invention discloses a multimodal intelligent cost accounting method and system for castings. It features adaptive processing of multimodal drawings: for the first time, it achieves unified automated parsing of planar drawings (two-dimensional annotations) and three-dimensional drawings (PMI information). Image recognition and rule completion solve the problem of incomplete information in planar drawings, breaking through the limitation of traditional systems that only support a single drawing type. Furthermore, it deeply binds PMI information to the process: casting process constraints extracted from the three-dimensional drawings (such as draft angle, dimensional tolerances, surface roughness, and flaw detection levels) are directly used as core parameters for process matching. The 2D drawings indirectly map process requirements through feature recognition, achieving a quantitative correlation between "drawing information - process parameters - cost"; the process model's parameters self-learn: dynamically optimize parameters such as material loss rate and process time based on historical production data, upgrading cost accounting from "theoretical value calculation" to "actual production adaptation", with a deviation rate of ≤3%; full-process automation and dynamic market adaptation: from drawing input to quotation output, no manual intervention is required (efficiency improved by more than 10 times), and real-time market data (material prices, regional costs) and order variables (batch size, lead time) are integrated, resulting in quotations that are both accurate and competitive in the market. Attached Figure Description

[0014] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0015] Figure 1 A flowchart of the method provided by the present invention; Figure 2 The structural block diagram provided for this invention. Detailed Implementation

[0016] 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.

[0017] This invention discloses a multimodal intelligent cost accounting method for castings, the specific steps of which include: S1. Receive 2D or 3D drawings, extract and structure the process information, and automatically supplement the missing information in the 2D drawings; S2. Based on the structured process information, match the target casting process route and key process parameters; S3. Calculate the basic total cost based on the target casting process route, including material cost, process cost and additional cost; S4. By combining historical data deviations, real-time market data, and order variables, the basic total cost is corrected to obtain the corrected total cost; S5. Generate a quotation result including cost details and process description based on the revised total cost.

[0018] To further optimize the above technical solution, the specific steps for obtaining and preprocessing the design drawings of the castings are as follows: Drawing reception: Supports users to submit drawings via local upload and cloud synchronization, and is compatible with formats including 2DCAD and 3D models; Provides users with convenient and diverse data submission entry points to ensure smooth data entry into the system. Offers web or client interfaces, allowing users to upload files via drag-and-drop or click. Cloud synchronization: Integrates with third-party cloud storage services (such as Alibaba Cloud OSS, Baidu Cloud Drive, Google Drive APIs, etc.). After user authorization, the system automatically retrieves drawing files from the specified directory.

[0019] Unified format: Non-standard format drawings are converted into an intermediate format that the system can parse through an open-source CAD kernel; The system standardizes the format to solve the "data silo" problem by converting heterogeneous data from different CAD software and versions into a standard format that can be uniformly processed and parsed within the system. The system calls the selected kernel library, reads the original file, reconstructs its geometry in memory, and then exports it to the specified intermediate format (such as STEP). For 2D drawings (DXF), it extracts information such as lines and blocks.

[0020] Noise removal: The computer vision edge detection algorithm (Canny operator) is used to remove non-geometric information such as annotation lines, annotation text, and reference symbols in the drawings, and only retain the outline data of the casting body and the gating system. By stripping away auxiliary information unrelated to casting formation, the contour data of the casting body and gating system are accurately extracted, providing clean geometric input for subsequent simulation analysis (such as filling and solidification simulation). Coordinate calibration: Establish a three-dimensional world coordinate system with the center of the bottom surface of the casting as the origin, the length direction as the X-axis, the width direction as the Y-axis, and the height direction as the Z-axis; Establishing a unified and standardized world coordinate system to ensure that all castings being analyzed are under the same spatial reference is a prerequisite for accurate simulation (such as setting the direction of gravity in gravity casting) and automated processing. First, identify the reference plane: automatically identify the largest plane of the casting using an algorithm, or specify the plane with the smallest Z-axis coordinate, and determine it as the "bottom plane"; second, calculate the origin: find the center point of the outer contour rectangle of the "bottom plane" and set this point as the new coordinate origin (0,0,0); third, align the axes: align the normal vector direction of the bottom plane to the Z-axis; calculate the principal direction of the casting on the bottom plane (usually using PCA principal component analysis or selecting the direction of the longest side of the bottom plane), and align it to the X-axis; the Y-axis is automatically determined by the cross product of the Z-axis and the X-axis (Y=Z×X); finally, transform the matrix: based on the above calculations, generate a transformation matrix, and perform rotation and translation operations on the entire model (including the casting and gating system) to reposition it to the standard coordinate system.

[0021] To further optimize the above technical solution, the core features of the casting are analyzed. Through "geometric feature extraction algorithm + drawing attribute recognition", three types of key pricing features are extracted from the preprocessed drawings: geometric features, material features, and process features. Among them, geometric features include: casting body volume, gating and riser volume, minimum / average wall thickness, maximum outline size, number / diameter / depth of holes and slots; material features include: casting material, material density, melting point, and tensile strength; process features include: casting precision grade, surface roughness, heat treatment requirements, and non-destructive testing requirements.

[0022] Specifically, the voxelization algorithm calculates the volume, accurately calculating the volume of the casting body and the gating system, providing basic data for calculating material costs, designing gating systems (such as ladle size), and conducting solidification simulations.

[0023] Furthermore, a voxel is a pixel in three-dimensional space, and voxelization is the process of discretizing a continuous geometric model into a three-dimensional mesh composed of tiny cubes (voxels).

[0024] Create a bounding box, using a cube (AABB axis bounding box) that can completely enclose the 3D model as the computational space.

[0025] Mesh generation: Divide the large cube into Nx*Ny*Nz smaller cubes (voxels). The accuracy (resolution) of the division is determined by the size of the voxels. The smaller the voxels, the higher the accuracy, but the greater the computational cost.

[0026] For each voxel, determine whether its center point is inside or outside the model. Common methods include the ray casting method or the winding number algorithm.

[0027] Volume calculation: Count the number of voxels C marked as "inside", and the total volume of the model V = C * (voxel size). 3 The wall thickness analysis algorithm (based on distance transformation) extracts the wall thickness, automatically and accurately measures the wall thickness of various parts of the casting, and identifies thick hot spots (areas prone to shrinkage cavities) and excessively thin areas (areas prone to incomplete filling). The specific steps are as follows: Input: The binary voxel mesh generated in the above steps (1 represents the interior of the material, 0 represents the exterior).

[0028] Perform distance transformation: Traverse the entire mesh using algorithms such as Chamfer or Euclidean distance transformation (EDT).

[0029] Extracting wall thickness: For any point inside the model, twice the distance d (2d) is the local wall thickness at that point. (Because distance d is the distance to one side of the surface, and the distance to the opposite side of the surface is also d).

[0030] Visualization and Analysis: By mapping distance values ​​to colors, an intuitive wall thickness cloud map can be generated. The maximum, minimum, and average wall thickness can be calculated, and a report can be generated to identify hot spot areas.

[0031] For 2D CAD drawings: Volume calculation using the profile extrusion method. When the user only uploads a 2D drawing (such as DXF / DWG), 3D voxelization analysis cannot be performed directly. The specific steps are as follows: Obtaining Closed Profiles: After the "Noise Removal" step, the system should have obtained one or more clean closed profiles representing the cross-section of the casting.

[0032] Calculate the cross-sectional area: Use the Shoelace formula (Gaussian area formula) to calculate the area A enclosed by each closed contour line.

[0033] Specifying the stretch length: This requires the user to provide a parameter—the length (L) of the casting in the direction perpendicular to the drawing plane. This is usually the main thickness direction of the casting.

[0034] Volume calculation: Volume V = Area(A) * Length(L).

[0035] Material characteristics are obtained from drawing attribute fields (such as "Material: A356") or technical requirement annotations. If they cannot be identified, the user is prompted to supplement them. Process characteristics are analyzed by parsing the technical requirements text of the drawings using natural language processing (NLP) and matching it with a preset process keyword library (such as "T6" and "UT inspection").

[0036] To further optimize the above technical solution, a preset "casting process rule library" is invoked, and the rule engine is used to automatically match the process with the mold solution. The specific steps are as follows: The arbitrary combination of input feature sets is mapped to one or more suitable core casting processes through predefined logical rules; After the core process is determined, the system further calls the corresponding auxiliary process schemes and specific parameters from the process knowledge base according to the finer-grained technical requirements. Retrieve mold information corresponding to the core casting process and auxiliary process from the "Mold Parameter Library".

[0037] Specifically, the input feature set of any combination is mapped to one or more suitable core casting processes through predefined logical rules. The rule engine pre-stores the "feature-process" mapping logic, as shown in the following example: If the material is aluminum alloy / zinc alloy, the minimum wall thickness is ≤5mm, and the batch size is ≥1000 pieces, then match the "die casting process". If "material is cast steel / cast iron + volume ≥ 1m and accuracy requirement CT10" → match "sand casting process"; 3 + If "material is high-temperature alloy + precision requirement CT7 + surface roughness Ra≤3.2μm" → match "investment casting process"; (The process rule library supports manual addition / modification. For example, when a company adds a "lost foam casting" process, it can add the corresponding feature matching rules.)

[0038] After the core process is determined, the system further calls the corresponding auxiliary process schemes and specific parameters from the process knowledge base according to the finer-grained technical requirements. Matching process based on heat treatment requirements: such as "T6 aging" → "solution (530℃×2h) + water quenching + aging (120℃×4h)"; Cleaning processes based on surface roughness matching: such as "Ra≤6.3μm" → "shot peening + grinding"; Based on the requirements of non-destructive testing, the testing process is matched: such as "100% UT testing" → "full-dimensional ultrasonic testing after casting cleaning".

[0039] Retrieving mold information for the corresponding process from the "Mold Parameter Library" is shown in the following example: Die casting process → Matching "H13 steel metal mold", pre-stored parameters include: mold manufacturing cost (100,000-500,000 RMB, related to casting size), estimated life (50,000-100,000 pieces, related to material), maintenance cost (5% of the total mold price per year). Sand casting → Matching "resin sand mold", pre-stored parameters include: single mold cost (500-2000 yuan, depending on the complexity of the casting) and lifespan (1 piece / set, one-time consumption); Investment casting → Matching "wax mold", with pre-stored parameters including: mold manufacturing cost (50,000-150,000 RMB) and lifespan (20,000-50,000 pieces).

[0040] To further optimize the above technical solution, the specific steps for calculating the total cost of casting and generating the initial cost estimate are as follows: Based on the "casting cost parameter library", the total cost of the casting is calculated item by item as follows: Material cost is the core cost item for castings, and it needs to take into account "cast body + gating and riser + process losses", as shown in the following formula: Total material cost = (mass of casting body + mass of gating and riser) × unit price of material × (1 + material loss rate); Mass of the casting body = Volume of the casting body × Density of the material; The mass of the gating and riser = the volume of the gating and riser × the density of the material (if the drawings do not include a gating and riser design, it is calculated by default as 10%-15% of the mass of the casting body, which can be manually adjusted). Material unit price: retrieved from the "Real-time Material Price Database" (updated monthly, including VAT), such as A356 aluminum alloy 22 yuan / kg, HT250 gray cast iron 4.5 yuan / kg; Material loss rate: Preset according to process (die casting 3%-5%, sand casting 8%-12%, investment casting 5%-8%), and support enterprises to calibrate according to actual production data.

[0041] Different allocation methods are used depending on the mold type (disposable / reusable): One-time molds (such as sand molds): Unit cost of mold = Cost of mold per piece (no allocation, fully included in the cost of a single casting); Reusable molds (such as die-casting / wax molds): Unit cost of mold = (mold manufacturing cost + mold maintenance cost × estimated service life) ÷ estimated lifespan of mold; (Example: A die-casting mold worth 150,000 yuan, with a lifespan of 100,000 pieces, a maintenance cost of 7,500 yuan / year, and an estimated lifespan of 3 years → Unit mold cost = (150,000 + 7,500 × 3) ÷ 100,000 ≈ 1.73 yuan / piece).

[0042] Labor costs cover the entire casting process and are calculated by adding up the labor hours of each process step multiplied by the corresponding fee rate, as shown in the following formula: Total labor cost = Molding labor cost + Smelting labor cost + Casting labor cost + Cleaning labor cost + Heat treatment labor cost; The calculation logic for working hours at each stage is as follows: Molding time = Number of castings required for molding ÷ Molding capacity (e.g., 80 molds / hour for medium-sized horizontal molding line with clay sand, 200 molds / hour for vertical molding line with clay sand) Melting time = (cast body mass + gating and riser mass) ÷ smelting equipment capacity (e.g., die casting smelting furnace capacity 150kg / h, sand casting cupola furnace capacity 500kg / h). Pouring time = Number of casting cavities × Pouring time per mold (based on process preset: die casting 0.1h / mold, sand casting 0.5h / mold); Cleaning time = casting complexity coefficient × casting body mass ÷ cleaning equipment capacity (complexity coefficient: simple part 1.0, medium complexity 1.2, complex part 1.5, automatically determined based on the number of holes / grooves / wall thickness distribution); Heat treatment time = heat treatment furnace load ÷ casting body mass × holding time (e.g., T6 aging holding for 4 hours, heat treatment furnace load 50kg / furnace). Labor hour rates: Preset according to process steps (e.g., 80 yuan / h for smelting, 60 yuan / h for casting, 50 yuan / h for cleaning, and 100 yuan / h for heat treatment), and support enterprises to update according to local labor costs.

[0043] Ancillary costs include energy consumption, testing, and other expenses, as detailed below: Energy consumption cost = Σ (power of equipment in each process step × labor hours × electricity price) (e.g., smelting furnace power 200kW, heat treatment furnace power 50kW, electricity price 1.2 yuan / kWh); Testing cost = Unit price of testing type × Testing ratio (e.g., UT testing is 50 yuan / piece, MT testing is 30 yuan / piece; 100% testing is fully included, and sampling testing is calculated proportionally). Other costs: including the cost of consumables such as release agent and coating, estimated at 1%-2% of the weight of the casting (e.g., 0.5 yuan / kg).

[0044] Initial cost estimate = material cost + mold cost + labor cost + auxiliary cost.

[0045] The initial pricing results are dynamically optimized based on market fluctuations, batch discounts, process complexity, and manual fine-tuning. The specific steps are as follows: Market volatility correction: Adjust material costs and electricity prices based on the "monthly market volatility coefficient library" (e.g., if the price of aluminum alloy increases by 3% in a given month, then the material cost will be multiplied by 1.03). Batch discount adjustment: Set the discount coefficient according to the order batch (0.98 discount for 100-499 items, 0.95 discount for 500-999 items, and 0.9 discount for ≥1000 items). The batch information is entered by the user or synchronized from the order system. Process complexity correction: For castings with complex structures (such as irregular curved surfaces and deep holes), adjust the time cost according to the complexity coefficient (1.05-1.2) (the complexity is automatically determined based on the surface curvature change rate).

[0046] Manual fine-tuning (optional): Users can manually adjust key parameters (such as material loss rate and mold life) in the system interface. After adjustment, the system updates the pricing results in real time and records the parameter modification log (for easy traceability).

[0047] Output results: Generates standardized cost breakdown reports, including "cost item name, calculation basis, amount, and percentage" (example: material cost 252 yuan, accounting for 53%; mold cost 1.7 yuan, accounting for 0.4%), supports export in PDF / Excel format, and generates a visual cost composition pie chart (to facilitate enterprises and customers in analyzing cost optimization potential).

[0048] In another embodiment, a multimodal intelligent cost accounting system for castings, applying the above method, includes: The drawing acquisition unit is used to receive 2D / 3D drawings, achieving format compatibility and cloud synchronization, and includes a format conversion module and a cloud interaction module. The drawing parsing unit is used to extract geometric / material / process features, and includes a geometric feature extraction module, a material feature recognition module, and a process feature parsing module. The process matching unit is used to match casting processes with mold solutions, and includes a casting process library, a rule engine, and a mold parameter library. The cost calculation unit is used to calculate costs item by item, including material cost module, mold cost module, labor cost module, and auxiliary cost module; The pricing optimization unit is used to dynamically correct pricing results, and includes a market volatility coefficient library, a batch discount module, and a manual adjustment module. The storage unit is used to store drawings, process parameters, and pricing results, and includes a drawing storage module, a process parameter storage module, and a pricing result storage module. The interactive unit is used to display and output pricing results, including a report generation module, a visualization interface module, and a historical query module.

[0049] To further optimize the above technical solutions, the casting process library pre-stores process parameters for sand casting, die casting, and investment casting, including material loss rate, smelting equipment capacity, and labor cost rate, and supports enterprises to update them online through administrator accounts.

[0050] To further optimize the above technical solution, the interactive unit supports access from both web and PC. The generated cost breakdown report includes the cost item name, calculation basis, amount, and percentage. It supports export in PDF / Excel format and generates a visual chart of cost composition.

[0051] The following example, “Aluminum alloy housing casting for automotive gearbox” (3D model format .step, order batch 1000 pieces), illustrates the implementation process of this invention and verifies the feasibility of the technical solution.

[0052] Implementation prerequisites (system preset data): Preset data for the process library:

[0053] Mold parameter library preset data:

[0054] Preset data for cost parameter library: Material unit price: A356 aluminum alloy 22 yuan / kg (fluctuation +3% this month), density 2.7g / cm³; 3 Electricity price: 1.2 yuan / kWh; Testing unit price: UT testing 50 yuan / piece (100% testing); Consumable cost: 0.5 yuan / kg (weight of the casting body); Bulk discount: 0.9% off for orders of ≥1000 pieces.

[0055] Specific implementation steps: Step 1: Drawing Acquisition and Preprocessing; Users upload "Automotive Gearbox Aluminum Alloy Housing.step Model" via the Web interface. The system converts it to .brep format using the FreeCAD kernel, removes annotation lines (such as "M8 Threaded Hole" annotation) from the model, and establishes a three-dimensional coordinate system with the center of the bottom surface of the housing as the origin.

[0056] Step 2: Core Feature Analysis; Geometric Features: The volume of the shell body is calculated to be 3.5L (3500cm³) using a voxelization algorithm, and the volume of the gating system is 0.3L (300cm³); The minimum wall thickness is 3.5mm and the average wall thickness is 4.2mm using a wall thickness analysis algorithm; The maximum outline dimensions are 380×220×280mm; Material Features: The material is identified as A356 aluminum alloy with a density of 2.7g / cm³ from the model attribute field "Material: A356AluminumAlloy"; Process Features: The NLP analysis technology requires "CastingTolerance: CT8, SurfaceRoughness: Ra12.5, HeatTreatment: T6, NDT: 100%UT", and the accuracy is extracted as CT8, T6 heat treatment, and 100%UT inspection. 333 Step 3: Matching process and mold design; rule engine based on feature matching: Core process: Aluminum alloy + minimum wall thickness 3.5mm + batch size 1000 pieces → die casting process; Auxiliary processes: T6 heat treatment (solution treatment at 530℃ for 2 hours + water quenching + aging at 120℃ for 4 hours), UT testing (100%). Mold solution: Die casting mold (H13 steel, manufacturing cost 150,000 yuan, lifespan 100,000 pieces, maintenance cost 7,500 yuan / year, estimated lifespan 3 years).

[0057] Step 4: Cost Calculation 4.1 Material Costs: Total mass of casting = (3500 + 300) cm; 3 ×2.7g / cm 3 =10260g=10.26kg Total material cost = 10.26kg × 22 yuan / kg × (1 + 4%) × (1 + 3% market fluctuation) = 10.26 × 22 × 1.04 × 1.03 ≈ 246.8 yuan; 4.2 Mold Cost: Unit cost of mold = (150,000 yuan + 7,500 yuan / year × 3 years) ÷ 100,000 pieces = (150,000 + 22,500) ÷ 100,000 ≈ 1.73 yuan / piece; 4.3 Labor Costs: Smelting time = 10.26kg ÷ 150kg / h ≈ 0.068h, cost = 0.068 × 80 ≈ 5.4 yuan; Casting time = 1 type / piece × 0.1h / type = 0.1h, cost = 0.1 × 60 = 6 yuan; Cleaning time = 1.2 (shell contains 12 holes, complexity coefficient 1.2) × 10.26kg ÷ 50kg / h (cleaning capacity) ≈ 0.246h, cost = 0.246 × 50 ≈ 12.3 yuan; Heat treatment time = 50kg / furnace (heat treatment furnace loading) ÷ 10.26kg / piece × 4h (holding time) ≈ 1.95h, cost = 1.95 × 100 ≈ 195 yuan; Total labor cost = 5.4 + 6 + 12.3 + 195 ≈ 218.7 yuan; 4.4 Ancillary Costs: Energy consumption cost = (200kW × 0.068h + 50kW × 1.95h) × 1.2 yuan / kWh = (13.6 + 97.5) × 1.2 ≈ 133.3 yuan; Testing cost = 50 yuan / piece × 100% = 50 yuan; Consumable cost = 10.26kg × 0.5 yuan / kg ≈ 5.1 yuan; Auxiliary cost = 133.3 + 50 + 5.1 ≈ 188.4 yuan; 4.5 Initial pricing result = 246.8 + 1.73 + 218.7 + 188.4 ≈ 655.6 yuan.

[0058] Step 5: Cost Optimization and Output; Batch discount correction: 1000 pieces are eligible for a 0.9 discount, the corrected result = 655.6 × 0.9 ≈ 590.0 yuan; Manual fine-tuning: The user adjusted the material loss rate from 4% to 3.8% (actual production data of the enterprise). The system updated the material cost in real time to 245.2 yuan. The final pricing result = 245.2 + 1.73 + 218.7 + 188.4 ≈ 654.0 yuan × 0.9 ≈ 588.6 yuan; Output results: Generate a detailed cost breakdown report (as shown in Table 1), export it in PDF format, and generate a pie chart of cost composition (materials 41.7%, labor hours 37.2%, auxiliary materials 32.0%, molds 0.3%).

[0059] Table 1: Detailed Price List for Aluminum Alloy Housings of Automotive Transmissions The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section. The above description of the disclosed embodiments enables those skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the present invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A multimodal intelligent cost accounting method for castings, characterized in that, include: S1. Receive 2D or 3D drawings, extract and structure the process information, and automatically supplement the missing information in the 2D drawings; S2. Based on the structured process information, match the target casting process route and key process parameters; S3. Calculate the basic total cost based on the target casting process route, including material cost, process cost and additional cost; S4. By combining historical data deviations, real-time market data, and order variables, the basic total cost is corrected to obtain the corrected total cost; S5. Generate a quotation result including cost details and process description based on the revised total cost.

2. The multimodal intelligent cost accounting method for castings according to claim 1, characterized in that, The specific steps for S1 are as follows: If the input is a 3D drawing, the PMI information is parsed and extracted, and converted into structured parameters, including geometric dimensions, tolerances, materials, and process constraints. If the input is a 2D drawing, the OCR system recognizes the dimensions, process symbols, and material codes, and the missing parameters are completed using a rule engine. Based on the default process requirements of the materials, a structured parameter set is generated.

3. The multimodal intelligent cost accounting method for castings according to claim 1, characterized in that, The structured parameters of S1 are input into the process knowledge graph, and the optimal casting process route and key parameters are matched through a machine learning model. The specific steps are as follows: S201, Construct or update the process knowledge graph based on the input structured process parameters, and clarify the nodes and relationships; S202, Utilize graph features for data embedding and representation learning, and train a machine learning model in combination with historical process data to learn the complex mapping between process routes and parameters; S203, For new input parameters, Retrieve and recommend the optimal process route and key parameter combination in the graph through model reasoning and similarity matching; S204, Iteratively optimize the model and knowledge graph using actual production feedback data.

4. The multimodal intelligent cost accounting method for castings according to claim 1, characterized in that, The specific steps for S3 are as follows: The process-cost calculation model is invoked, and the total cost of the casting is calculated based on the S2 process route: material cost + mold cost + labor cost + auxiliary cost. The total material cost is calculated as follows: (mass of the casting body + mass of the gating and riser) × unit price of materials × (1 + material loss rate). Casting body mass = casting body volume × material density; Gating and riser mass = gating and riser volume × material density. If the drawing does not have a gating and riser design, it is calculated by default as 10%-15% of the casting body mass and manually adjusted; Material unit price: retrieved from the "real-time material price database"; Material loss rate: preset according to the process. Different allocation methods are used depending on the mold type (disposable / reusable): One-time mold: Unit cost of mold = Cost of a single mold piece; Reusable molds: Unit cost of mold = (Mold manufacturing cost + Mold maintenance cost × Estimated service life) ÷ Estimated lifespan of mold; Labor cost calculation covers the entire casting process and is calculated by summing "labor hours of each process step × corresponding cost rate of each process step", as shown in the following formula: Total labor cost = Molding labor cost + Smelting labor cost + Casting labor cost + Cleaning labor cost + Heat treatment labor cost; Molding time = Number of castings required for molding ÷ Molding capacity Smelting time = (mass of casting body + mass of gating and riser) ÷ capacity of smelting equipment; Pouring time = Number of casting cavities × Pouring time per mold; Cleaning time = casting complexity coefficient × casting body mass ÷ cleaning equipment capacity; Heat treatment time = heat treatment furnace loading capacity ÷ casting body mass × holding time; Labor cost rate: Preset according to process steps; Ancillary costs include energy consumption costs, testing costs, and other expenses, as detailed below: Energy cost = Σ (power of equipment in each process step × labor hours × electricity price); Testing cost = Unit price of testing type × Testing ratio; Other costs: including release agent and paint consumables, estimated at 1%-2% of the casting's mass.

5. The multimodal intelligent cost accounting method for castings according to claim 1, characterized in that, The specific steps for S4 are as follows: S401 identifies and quantifies deviations in historical cost data, and calculates the average deviation rate of each cost item as a basic correction coefficient by comparing historical forecasts with actual values. S402 incorporates real-time market data and dynamically adjusts cost items that are strongly correlated with the market through a preset sensitivity model; S403 performs customized cost increase / decrease calculations based on specific variables of the current order; S404 combines the aforementioned historical deviation correction, market dynamic adjustment, and order variable impact, and applies them to the base total cost through an integrated calibration model, outputting the corrected total cost.

6. A multimodal intelligent cost accounting system for castings, employing the multimodal intelligent cost accounting method for castings as described in any one of claims 1-6, characterized in that, include: The multi-source drawing parsing module is used to receive 2D or 3D drawings, extract and structure the process-related information in them, including dimensional parameters, tolerance requirements, material information, process constraints and quality requirements. Missing information in 2D drawings is automatically supplemented based on a database of similar products. The intelligent process matching module has a built-in casting process knowledge graph. Based on the structured process-related information, it matches the target casting process route and key process parameters through a machine learning model. The large-scale process-cost calculation model includes a material cost sub-model, a process cost sub-model, and an additional cost sub-model, which is used to calculate the basic total cost based on the target casting process route and structured process information. The big data dynamic optimization module connects to the historical quotation database and real-time market data to perform deviation calibration, market adaptation, and order variable adjustment on the basic total cost, and outputs the corrected total cost. The quotation generation module generates a quotation result that includes cost details, process descriptions, and price suggestions based on the revised total cost.

7. A multimodal intelligent cost accounting system for castings according to claim 6, characterized in that... include: Intelligent process matching module: It has a built-in casting process knowledge graph, which includes the applicable conditions of 12 types of processes such as clay sand casting, resin sand casting, lost foam casting, investment casting, and die casting. Based on the structured parameters output by the multi-source drawing parsing module, it matches the optimal process route through fuzzy reasoning and machine learning models.

Images