Mold cost calculation method, device, equipment and storage medium

Abstract

The embodiments of the present invention relate to the field of cost calculation, and disclose a mold cost calculation method, apparatus, equipment, and storage medium, including: acquiring multimodal data related to target mold manufacturing; extracting feature data related to target mold manufacturing cost from the multimodal data; extracting time series process data from the feature data based on a preset spatiotemporal graph network model, analyzing the first relationship between the time series process data and cost over time, and calculating the first cost according to the first relationship; and / or extracting spatial process data from the feature data based on the spatiotemporal graph network model, analyzing the second relationship between the spatial process data and cost, and calculating the second cost according to the second relationship. This application aggregates manufacturing data from different links, and uses a preset spatiotemporal graph network model to capture the first relationship between process parameters and cost over time, and the second relationship between the mutual correlation between different nodes and cost, thereby improving the accuracy of cost calculation.

Description

Technical Field

[0001] The embodiments of the present invention relate to the field of cost calculation, and in particular to a mold cost calculation method, apparatus, device and storage medium. Background Art

[0002] The mold industry plays a vital role in the manufacturing industry and is a key link in product formation. However, mold cost control faces multiple challenges, which seriously restricts the company's refined management and profitability.

[0003] The mold manufacturing process involves multiple steps, including design, processing, assembly, and testing. Data from each step is often independent, creating information silos. Industry surveys indicate that data utilization across all manufacturing factors is insufficient. This fragmented data makes it difficult for companies to comprehensively and systematically understand cost structures and trends, hindering refined cost management. Summary of the Invention

[0004] The purpose of the present invention is to at least provide a mold cost calculation method, device, equipment and storage medium, which can at least solve the technical problems of insufficient utilization of all manufacturing factor data in the mold manufacturing process, difficulty in comprehensively and systematically grasping the cost structure and change rules, and realizing refined cost management, and at least achieve the effect of more refined calculation of mold manufacturing costs.

[0005] To solve the above technical problems, at least one embodiment of the present application provides a mold cost calculation method, comprising: acquiring multimodal data related to target mold manufacturing;

[0006] extracting feature data related to the target mold manufacturing cost from the multimodal data;

[0007] Extracting time series process data from the feature data based on a preset spatiotemporal graph network model, analyzing a first relationship between the time series process data and the cost, and calculating a first cost of the target mold based on the first relationship; and / or

[0008] Based on the spatiotemporal graph network model, spatial process data is extracted from the feature data, and a second relationship between the spatial process data and the cost is analyzed, and the second cost of the target mold is calculated according to the second relationship, wherein the spatial process data includes interrelated process parameters and / or interrelated equipment.

[0009] At least one embodiment of the present application further provides a mold cost calculation device, comprising:

[0010] A data acquisition module, used to acquire multimodal data related to target mold manufacturing;

[0011] a feature data extraction module, configured to extract feature data related to the target mold manufacturing cost from the multimodal data;

[0012] a first cost calculation module, configured to extract time series process data from the feature data based on a preset spatiotemporal graph network model, analyze a first relationship between the time series process data and the cost, and calculate a first cost of the target mold based on the first relationship; and / or

[0013] A second cost calculation module is used to extract spatial process data from the feature data based on the spatiotemporal graph network model, and analyze a second relationship between the spatial process data and the cost, and calculate the second cost of the target mold according to the second relationship, wherein the spatial process data includes interrelated process parameters and / or interrelated equipment.

[0014] At least one embodiment of the present application also provides an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the above-mentioned mold cost calculation method.

[0015] At least one embodiment of the present application further provides a computer-readable storage medium storing a computer program, which implements the above-mentioned mold cost calculation method when executed by a processor.

[0016] The mold cost calculation method, apparatus, equipment, and storage medium provided in the embodiments of the present application acquire multimodal data related to target mold manufacturing: actively collect heterogeneous data (such as design drawings, processing procedures, processing parameters, etc.) covering multiple links such as design, processing, assembly, and testing, and bring together data originally scattered across different departments, systems, and links, breaking down physical and logical "information silos" at the source and providing a data foundation for subsequent comprehensive analysis; identify and extract key features that are strongly correlated with cost, thereby improving the "value density" and targetedness of the data; address the pain point of "difficulty in fully grasping the cost structure and change patterns", utilize a preset spatiotemporal graph network model to capture the change patterns of process parameters, equipment status, resource consumption, etc. over time, analyze the primary relationship between the change patterns and costs, and solve the problem of difficulty in capturing dynamic change patterns in the cost structure; and / or analyze the secondary relationship between the mutual correlations between different nodes and costs, such as how design complexity affects the selection of processing parameters, how processing accuracy affects assembly difficulty and rework rate, etc., to quantify how a design decision affects the costs of all subsequent links through a chain reaction. The final cost calculation results in primary cost, secondary cost, or a combination of these. Rather than simply adding up the costs of individual links, they are refined calculations that integrate the dynamic interactions of the entire process and all factors. This allows enterprises to comprehensively and systematically grasp the true cost structure and achieve refined cost management.

[0017] In some optional embodiments, calculating the first cost of the target mold according to the first relationship, and calculating the second cost of the target mold according to the second relationship, includes:

[0018] Constructing a process-cost association matrix based on the first relationship and the second relationship, wherein the process-cost association matrix is used to quantitatively analyze the degree and direction of influence of characteristic data on cost;

[0019] The first cost and the second cost are calculated according to the process-cost association matrix.

[0020] In this embodiment, the matrix integrates multi-dimensional features such as design, process, equipment, and environment, and captures the interaction effects between features through matrix operations. When the process is adjusted (such as changing materials or using new equipment), the matrix parameters are dynamically updated to instantly generate a cost forecast for the new solution.

[0021] In some optional embodiments, constructing a process-cost association matrix based on the first relationship and the second relationship includes: using the space-time graph network model to abstract each of the feature data and cost elements into nodes in the space-time graph structure, the cost elements including at least one of raw material procurement cost, processing cost and processing loss cost; analyzing the historical sequence data of each node through the timing model in the space-time graph network model to obtain the first relationship between the feature data and the cost element; aggregating the information of neighboring nodes through the graph neural network model in the space-time graph network model to obtain the second relationship between the feature data and the cost element; constructing the process-cost association matrix based on the first relationship and the second relationship.

[0022] By abstracting characteristic data and cost elements into nodes within a spatiotemporal graph structure using a spatiotemporal graph network model, the previously complex and disorganized characteristic data and cost element information can be presented in an intuitive and structured manner. This structured representation facilitates subsequent data analysis and processing, allowing complex process-cost relationships to be presented in a clearer graph structure, making them easier to understand and manipulate.

[0023] In some optional embodiments, the method further includes:

[0024] Extracting structured data from the feature data, wherein the type of the structured data includes at least one of geometric complexity, process path, and material properties;

[0025] The third cost of the target mold is calculated according to the contribution of the structured data to the cost.

[0026] In this embodiment, structured data is static cost-related indicator parameters with a clear format and definition. Data types include geometric complexity, process paths, and material properties. This data typically exists in standardized tables or database fields, with each data item having a clear meaning and value range. Structured data provides a benchmark for cost calculation based on the mold's essential characteristics, making mold cost calculation results more comprehensive.

[0027] In some optional embodiments, extracting structured data from the feature data includes at least one of the following:

[0028] Extracting the geometric complexity through 3D point cloud curvature analysis; and

[0029] Extracting relevant parameters under the process path based on a numerical control machining instruction library, wherein the relevant parameters include at least one of path length, feed speed and cutting depth; and

[0030] The material properties are obtained by dynamic hardness map analysis, wherein the dynamic hardness map is used to simulate the change of material hardness with position, temperature, time or strain rate under actual processing conditions.

[0031] In this embodiment, the geometric complexity features are extracted through 3D point cloud curvature analysis to accurately quantify the mold geometric features, reflecting the complexity of the mold shape and its impact on processing difficulty and cost; the CNC machining instruction library contains rich processing information, and by extracting relevant parameters under the process path, such as path length, feed speed and cutting depth, etc., the process details of the mold manufacturing process can be fully understood; the dynamic hardness map can simulate the changes in material hardness with position, temperature, time or strain rate under actual processing conditions. The material properties obtained through dynamic hardness map analysis are closer to the actual performance of the material in the actual processing process, providing more accurate material property data for cost calculation and process design.

[0032] In some optional embodiments, the method further includes:

[0033] A weighted sum is taken for the first cost, the second cost, and the third cost to obtain a comprehensive cost for manufacturing the target mold.

[0034] In this embodiment, the first cost is calculated based on the relationship between time-series process data and cost, reflecting the impact of process changes over time on cost. The second cost is calculated based on the relationship between spatial process data and cost, reflecting the impact of spatial correlations between process parameters and equipment on cost. The third cost is calculated based on the contribution of structured data (such as geometric complexity, process path, and material properties) to cost, taking into account the impact of the mold's inherent properties on cost. Through weighted summation, these three different dimensions of cost information are combined, avoiding the one-sidedness that can result from single-dimensional cost calculations. This comprehensive cost more comprehensively reflects the various cost factors in the mold manufacturing process.

[0035] In some optional embodiments, the calculation formula of the comprehensive cost of the target mold includes the following:

[0036]

[0037] in, The comprehensive cost of the target mold, 、 and Represent the prediction functions based on XGBoost, LSTM and graph convolutional network respectively, For the structured data, is the timing process data, is the spatial process data, 、 and is the gating coefficient, which is dynamically adjusted through the preset attention mechanism.

[0038] In this embodiment, the attention mechanism can dynamically adjust the gating coefficients (i.e., weights) of XGBoost, LSTM, and GCN based on the characteristics of the input data. This dynamic adjustment enables the model to automatically assign the importance of different models in different situations. For example, when structured data has a greater impact on cost, the attention mechanism will increase the gating coefficient of XGBoost; when changes in time-series process data have a significant impact on cost, the gating coefficient of LSTM will be increased. This allows the method of the present application to adapt to complex and changing manufacturing environments, such as equipment status, material supply, process adjustments, etc. By dynamically adjusting the gating coefficients through the attention mechanism, the comprehensive cost calculation method can adapt to these complex and changing environments, adjust the model weights in real time, and ensure that the cost forecast always maintains a high level of accuracy.

[0039] In some optional embodiments, after obtaining the comprehensive cost of manufacturing the target mold, the method further includes:

[0040] Based on the comprehensive cost and real-time production line data, a cost control strategy is calculated using a preset reinforcement learning algorithm, wherein the cost control strategy includes at least one of process parameter setting, material selection, and processing path optimization, and the reward function of the reinforcement learning algorithm includes the degree of cost reduction and process quality;

[0041] The cost control strategy is verified using a digital twin model of a preset production line, and the cost control strategy is adjusted according to the verification results until the verification results meet the preset standards to obtain the final cost control strategy.

[0042] In this embodiment, the production line's digital twin model is a virtual reflection of the actual production line, allowing cost control strategies to be validated without disrupting actual production. By simulating the implementation of strategies within the digital twin model, potential issues and risks, such as equipment conflicts and process mismatches, can be identified in advance, thus avoiding adverse consequences such as increased costs, reduced quality, or production interruptions in actual production.

[0043] In some optional embodiments, extracting feature data related to the target mold manufacturing cost from the multimodal data includes:

[0044] Extracting initial feature data related to cost calculation based on the multimodal data;

[0045] The initial feature data is subjected to feature dimension compression to obtain the feature data finally related to cost calculation.

[0046] This example summarizes that high-dimensional data is difficult to visualize directly. However, by reducing the data to a two- or three-dimensional space through dimensional compression, the relationships and distribution of the data can be more intuitively displayed. For example, visualization tools such as scatter plots and line graphs can be used to display the relationship between compressed feature data and costs, helping to better understand the factors affecting costs. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] One or more embodiments are exemplarily described by the figures in the corresponding drawings, and these exemplified descriptions do not constitute limitations on the embodiments.

[0048] Figure 1 is a flow chart of a mold cost calculation method provided by an embodiment of the present application;

[0049] Figure 2 This is a schematic diagram of the structure of a mold cost calculation system provided by an embodiment of the present application;

[0050] Figure 3 This is a schematic diagram of the optimization effect of a mold cost calculation system provided by another embodiment of the present application;

[0051] Figure 4 This is a schematic diagram of a mold cost calculation device provided in another embodiment of the present application. DETAILED DESCRIPTION

[0052] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, each embodiment of the present application will be described in detail below with reference to the accompanying drawings. However, it will be understood by those skilled in the art that in each embodiment of the present application, many technical details are proposed to enable the reader to better understand the present application. However, even without these technical details and various changes and modifications based on the following embodiments, the technical solutions claimed in the present application can be implemented. The division of the following embodiments is for convenience of description and should not constitute any limitation on the specific implementation of the present application. The various embodiments can be combined and referenced with each other under the premise of no contradiction.

[0053] The mold industry plays a vital role in the manufacturing industry and is a key link in product formation. However, mold cost control faces multiple challenges, which seriously restrict companies' refined management and profitability. Through in-depth research on the current status of the industry, we found that there are three core difficulties in mold cost control:

[0054] First, the industry generally faces a serious dilemma of relying on experience. According to data released by the China Mold Industry Association in 2023, 78% of companies still use manual experience to estimate mold costs. This reliance on manual experience is not only inefficient but also inaccurate, easily leading to significant deviations between cost estimates and actual costs. Many companies report that mold cost amortization is difficult because molds are a component of the cost of producing product components, while typical cost accounting is based on finished or semi-finished products, which naturally makes mold cost amortization difficult. This dilemma results in companies lacking accurate cost information to support decision-making, making it difficult to conduct scientific cost management and optimization.

[0055] Secondly, insufficient dynamic responsiveness has become a key factor hindering the effectiveness of mold cost control. Traditional cost control systems often rely on batch processing and periodic adjustments, resulting in a typical system adjustment lag of 4-6 hours. In a rapidly changing market environment, this lag makes it difficult for companies to promptly respond to external factors such as raw material price fluctuations and shifts in market demand that impact costs. Especially in the current context of rising raw material prices, cost control has become a core element of mold and die enterprise management. Therefore, establishing a cost control system capable of real-time perception and rapid response has become a pressing issue for the industry.

[0056] Third, data silos severely restrict the depth and breadth of mold cost management. The mold manufacturing process involves multiple steps, including design, processing, assembly, and testing. Data from each step is often independent, creating information silos. Industry surveys indicate insufficient utilization of all manufacturing factor data. This fragmented data makes it difficult for companies to comprehensively and systematically understand cost structures and trends, hindering refined cost management. Calculating original mold costs requires significant manual data collection and duplication of effort, increasing management costs and reducing decision-making efficiency.

[0057] In order to solve the above-mentioned technical problem of insufficient accuracy in mold cost calculation, the present invention proposes a mold cost calculation method. The implementation details of the mold cost calculation method of this embodiment are specifically described below. The following content is only the implementation details provided for easy understanding and is not necessary for implementing this solution.

[0058] Example 1:

[0059] The mold cost calculation method of this embodiment can be applied to electronic devices with communication, computing and data storage capabilities, such as Figure 1 As shown, it includes:

[0060] Step 110, acquiring multimodal data related to target mold manufacturing;

[0061] In this embodiment, multimodal data refers to multi-source heterogeneous data in the target mold manufacturing process, which can be obtained from multiple manufacturing-related departments. For example:

[0062] Design department: Obtain geometric information from target mold design drawings, 3D models, and other materials, such as mold size, shape complexity, and structural features. This information directly affects the mold's processing difficulty and material usage, which in turn affects costs.

[0063] Production Department: Collect process parameter records during the production process, including machining parameters such as cutting speed, feed rate, and cutting depth, as well as equipment operating time, energy consumption, and maintenance records. Process parameters determine processing efficiency and product quality, while equipment operating conditions influence production efficiency and maintenance costs. In some optional embodiments, sensors are installed on production equipment to collect real-time operating parameters such as temperature, pressure, and vibration. This sensor data can provide more accurate, real-time equipment operating information, helping to promptly detect equipment failures and abnormalities, thereby reducing maintenance costs.

[0064] Procurement Department: Obtain the purchase price, supplier information, quality inspection reports, etc. of raw materials. The cost of raw materials is an important part of mold manufacturing costs. The price and quality of raw materials from different suppliers may vary. The quality inspection report can reflect whether the raw materials meet production requirements and avoid cost increases caused by raw material quality issues.

[0065] Quality Inspection Department: This department collects mold quality inspection data, such as dimensional accuracy, surface roughness, and hardness. This data can indicate whether the mold meets design requirements. If not, rework or scrapping may be necessary, increasing costs.

[0066] You can also query relevant historical and real-time data from the company's production management system, quality management system, procurement management system and other databases. The data in the database is usually organized and classified to facilitate quick access to the required information.

[0067] Step 120: extracting feature data related to the target mold manufacturing cost from the multimodal data;

[0068] In this embodiment, natural language processing technology can be used to extract keywords, phrases, and semantic information from text descriptions in design drawings and process documents, thereby obtaining feature data. Feature extraction is performed on the three-dimensional model of the target mold, the image of the design drawing, etc., and the shape features, contour features, texture features, etc. of the mold can be extracted. These feature data can intuitively reflect the geometric complexity and structural characteristics of the mold. Statistical features such as mean, variance, maximum, and minimum values are extracted from numerical data such as process parameter records and equipment operation data during the production process as feature data. These statistical features can reflect the distribution and changing trends of the data, which helps to analyze the relationship between process parameters and costs.

[0069] Step 130: extracting time series process data from the feature data based on a preset spatiotemporal graph network model, analyzing a first relationship between the time series process data and the cost, and calculating a first cost of the target mold based on the first relationship; and / or

[0070] Based on the spatiotemporal graph network model, spatial process data is extracted from the feature data, and a second relationship between the spatial process data and the cost is analyzed, and the second cost of the target mold is calculated according to the second relationship, wherein the spatial process data includes interrelated process parameters and / or interrelated equipment.

[0071] In this embodiment, the Space-Time Graph Network (STGCN) model is used to model and analyze the temporal and spatial relationships between feature data and costs. This pre-trained model can capture the complex relationships between feature data and costs. The STGCN model includes graph neural networks and time series models, combining the advantages of both to effectively capture the spatiotemporal dependencies in data. The details are as follows:

[0072] Time-series process data can be further subdivided into processing stage time-series data and equipment operation time-series data. Processing stage time-series data can reflect changes in process parameters and cost consumption in different processing stages. For example, different cutting parameters and processing times in the roughing and finishing stages have different impacts on costs. Equipment operation time-series data can reflect the equipment's operating efficiency, energy consumption changes, and other conditions, helping to analyze the impact of equipment operation on costs. In addition to extracting the raw values of time-series process data, time-series features can also be extracted, such as trends, periodicity, and volatility of the time-series data. For example, by analyzing the time-series trend of processing time, it is possible to determine whether production efficiency is stable. By analyzing the periodic changes in equipment energy consumption, it is possible to optimize equipment operation plans and reduce energy costs.

[0073] When analyzing the first relationship between time-varying process data and cost, the spatiotemporal graph network model uses its internal memory units and gating mechanisms to remember key process parameter information over the past period of time, thereby extracting features that reflect process change trends, analyzing the impact of these features on cost, and obtaining the first relationship.

[0074] In some optional embodiments, the spatiotemporal graph network model also extracts the spatial characteristics of spatial process data, such as the distribution characteristics of process parameters, the layout characteristics of equipment, etc. For example, the spatial distribution of different process parameters in the mold processing process is analyzed to determine whether there is a problem of excessive local processing difficulty leading to increased costs; analyze whether the layout of the equipment is reasonable and whether there is a situation where the material transportation distance is too long, leading to increased costs. For the interrelated process parameters and equipment in the spatial process data, the spatiotemporal graph network model can establish a spatial correlation relationship model. The graph structure can more intuitively display the correlation relationship of spatial process data, thereby analyzing and obtaining the second relationship between spatial process data and cost.

[0075] This embodiment changes the traditional cost management model, which primarily relies on qualitative analysis, and provides a dynamic cost control model. Specifically, by analyzing the impact of time-series process data and spatial process data on costs, it models and analyzes the complex relationship between feature data and costs. This effectively captures the cost impact of process parameters that vary over time and space, providing a decision-making basis for dynamic cost optimization.

[0076] In actual use, you can analyze only the first cost generated by the time-series process data, or only the second cost corresponding to the spatial process data, or you can combine the first and second costs for comprehensive calculation. This depends on actual production needs and is not limited here.

[0077] This embodiment acquires multimodal data related to target mold manufacturing: actively collects heterogeneous data (such as design drawings, processing procedures, and processing parameters) covering multiple links such as design, processing, assembly, and testing, and brings together data originally scattered across different departments, systems, and links, breaking down physical and logical "information silos" at the source and providing a data foundation for subsequent comprehensive analysis; identifies and extracts key features that are strongly correlated with cost, thereby improving the "value density" and targetedness of the data; addresses the pain point of "difficulty in fully grasping the cost structure and changing patterns" by utilizing a preset spatiotemporal graph network model to capture the changing patterns of process parameters, equipment status, resource consumption, etc. over time, analyzes the primary relationship between the changing patterns and costs, and solves the problem of difficulty in capturing dynamic changing patterns in the cost structure; and / or analyzes the secondary relationship between the mutual correlations between different nodes and costs, such as how design complexity affects the selection of processing parameters and how processing accuracy affects assembly difficulty and rework rate, thereby quantifying how a design decision affects the costs of all subsequent links through a chain reaction. The final cost calculation results in primary cost, secondary cost, or a combination of these. Rather than simply adding up the costs of individual links, they are refined calculations that integrate the dynamic interactions of the entire process and all factors. This allows enterprises to comprehensively and systematically grasp the true cost structure and achieve refined cost management.

[0078] In some optional embodiments, the first cost of the target mold is calculated based on the first relationship, and the second cost of the target mold is calculated based on the second relationship, including: constructing a process-cost association matrix based on the first relationship and the second relationship, the process-cost association matrix being used to quantitatively analyze the degree and direction of influence of characteristic data on cost; and calculating the first cost and the second cost based on the process-cost association matrix.

[0079] In this embodiment, the process-cost correlation matrix integrates multi-dimensional features such as design, process, equipment, and environment, and captures the interaction effects between features through matrix operations. For example, in this embodiment, the process-cost correlation matrix is a three-dimensional matrix, which includes:

[0080] The m-axis corresponds to process characteristics (such as injection pressure, cooling time, and material melting point);

[0081] The n-axis corresponds to cost factors (material cost, energy cost, and scrap loss);

[0082] t-axis, corresponds to the time window (capturing dynamic evolution);

[0083] A mathematical tool that encodes the triple relationship of "process characteristics-cost elements-time evolution" into a computable structure provides underlying support for dynamic cost control.

[0084] In some optional embodiments, constructing a process-cost association matrix based on the first relationship and the second relationship includes: using the space-time graph network model to abstract each of the feature data and cost elements into nodes in the space-time graph structure, the cost elements including at least one of raw material procurement cost, processing cost and processing loss cost; analyzing the historical sequence data of each node through the timing model in the space-time graph network model to obtain the first relationship between the feature data and the cost element; aggregating the information of neighboring nodes through the graph neural network model in the space-time graph network model to obtain the second relationship between the feature data and the cost element; constructing the process-cost association matrix based on the first relationship and the second relationship.

[0085] Specifically, the various characteristic data that influence costs during mold manufacturing are abstracted as nodes in a space-time graph. For example, characteristic data such as cutting speed, feed rate, cutting depth, equipment energy consumption, and raw material purchase price are each treated as a different node. Cost factors such as raw material purchase cost, processing cost, and processing loss cost are also abstracted as nodes in the space-time graph. Each cost factor node corresponds to a specific cost category. Edges between nodes represent the relationships between nodes, such as the synergistic relationship between process parameters and the causal relationship between characteristic data and cost factors.

[0086] When analyzing primary relationships, you can choose models such as long short-term memory networks (LSTMs), gated recurrent units (GRUs), or temporal convolutional networks (TCNs). These models can effectively capture long-term dependencies and short-term change patterns in time series data. For each feature data node and cost factor node, collect its historical sequence data. For example, historical data on cutting speed, monthly data on raw material purchase prices, quarterly data on processing costs, etc. Use a time series model to model and analyze the historical sequence data of each node, extracting features and patterns along the time dimension. By analyzing the correlation and causality between feature data nodes and cost factor nodes in the time series, determine the primary relationship between them. For example, how changes in cutting speed affect processing costs over time.

[0087] In a spatiotemporal graph, neighbor nodes can include spatial neighbors: multiple nodes physically connected / associated by process flows, and logical neighbors: multiple nodes whose data correlation meets a preset standard. For example, in the mold forming process, the stamping machine (molding equipment node) and the subsequent polishing machine (polishing equipment node) are continuous in the process flow, so the stamping machine and polishing machine are physically connected nodes. There is a strong data correlation between cutting speed (process parameter node) and mold surface roughness (quality indicator node). Generally speaking, excessively high cutting speeds may increase mold surface roughness, while excessively low cutting speeds may affect processing efficiency. When data analysis reveals that the correlation coefficient between cutting speed and surface roughness meets a preset standard (e.g., above 0.7), the surface roughness node becomes a logical neighbor of the cutting speed node.

[0088] When analyzing secondary relationships, you can choose models such as graph convolutional networks (GCNs), graph attention networks (GATs), or graph isomorphism networks (GINs). These models effectively aggregate information about neighboring nodes and capture the spatial relationships between them. By aggregating information about neighboring nodes, the spatial relationship between feature data nodes and cost factor nodes can be analyzed. For example, cutting speed and feed rate, as neighboring nodes, jointly influence the machining cost node, thus determining the secondary relationship between them.

[0089] Based on the first and second relationships, determine the degree and direction of influence of each feature data node on the cost factor node. Quantify the degree of influence into specific numerical values, such as correlation coefficients and weights, and indicate the direction of influence as positive or negative. For example, initialize an n×m matrix, where n is the number of feature data nodes and m is the number of cost factor nodes. Fill in the elements of the matrix based on the first and second relationships obtained from the analysis. If the cutting speed node has a degree of influence of 0.8 on the machining cost node and the direction of influence is positive, then enter +0.8 in the corresponding position in the matrix.

[0090] In this embodiment, the structure of the spatiotemporal graph network model (STGCN) is shown in the code:

[0091] ```Python

[0092] class STGCN(nn.Module):

[0093] def __init__(self):

[0094] self.gcn = GraphConv(in_dim=128, out_dim=64)

[0095] self.tcn = TemporalConv(64, 32)

[0096] def forward(self, graph_data):

[0097] spatial_feat = self.gcn(graph_data)

[0098] temporal_feat = self.tcn(spatial_feat)

[0099] return temporal_feat

[0100] ```

[0101] In this model, the GraphConv layer is used to capture spatial relationships, such as the mutual influence between different process parameters and the collaborative relationships between different equipment. The TemporalConv layer is used to capture temporal patterns, such as the trends in process parameters over time and the dynamic evolution of cost factors. By combining these two layers, STGCN is able to comprehensively and deeply analyze the complex relationship between process and cost. The STGCN model constructs a process-cost correlation matrix, quantifying the degree and direction of the impact of process parameters on cost. This matrix provides a clear direction and basis for cost optimization, guiding companies in targeted process improvements and cost control.

[0102] In this embodiment, the spatiotemporal graph network model is used to abstract the characteristic data and cost elements into nodes in a spatiotemporal graph structure. This allows for the presentation of previously complex and disorganized characteristic data and cost element information in an intuitive and structured manner. This structured representation facilitates subsequent data analysis and processing, allowing complex process-cost relationships to be presented in a clearer graph structure, making them easier to understand and operate.

[0103] In some optional embodiments, the method further includes: extracting structured data from the feature data, the type of the structured data including at least one of geometric complexity, process path, and material properties; and calculating the third cost of the target mold based on the contribution of the structured data to the cost.

[0104] In this embodiment, structured data is static cost-related indicator parameters with a clear format and definition. Data types include geometric complexity, process paths, and material properties. This data typically exists in standardized tables or database fields, with each data item having a clear meaning and value range. Structured data provides a benchmark for cost calculation based on the mold's essential characteristics, making mold cost calculation results more comprehensive.

[0105] In some optional embodiments, the extracting of structured data from the feature data includes at least one of the following: extracting the geometric complexity through 3D point cloud curvature analysis; and extracting relevant parameters under the process path based on a CNC machining instruction library, wherein the relevant parameters include at least one of path length, feed speed and cutting depth; and obtaining the material properties through dynamic hardness map analysis, wherein the dynamic hardness map is used to simulate the change of material hardness with position, temperature, time or strain rate under actual machining conditions.

[0106] In this embodiment, the geometric complexity features are extracted through 3D point cloud curvature analysis to accurately quantify the mold geometric features, reflecting the complexity of the mold shape and its impact on processing difficulty and cost; the CNC machining instruction library contains rich processing information, and by extracting relevant parameters under the process path, such as path length, feed speed and cutting depth, etc., the process details of the mold manufacturing process can be fully understood; the dynamic hardness map can simulate the changes in material hardness with position, temperature, time or strain rate under actual processing conditions. The material properties obtained through dynamic hardness map analysis are closer to the actual performance of the material in the actual processing process, providing more accurate material property data for cost calculation and process design.

[0107] These three types of features together constitute the feature space for mold cost calculation, covering the main influencing factors in the mold manufacturing process. Through dynamic feature engineering, the system can dynamically adjust feature weights and calculation methods based on real-time data and historical experience to adapt to different scenarios and changing needs. This dynamic nature makes the system highly adaptable and robust, able to cope with various changes and challenges in the mold manufacturing process. This is shown in Table 1 below:

[0108] Table 1

[0109]

[0110] Table 1 shows the structured data generated using 3D point cloud curvature analysis, G-code semantic parsing, and dynamic hardness mapping. The dimensions in this table refer to the dimensions used to calculate geometric complexity when analyzing set complexity, such as curvature and spatial flatness. The dimensions included in the process path include, for example, path length, feed rate, and depth of cut. Material characteristics can include hardness and heat resistance. Their contribution refers to their impact on cost. This structured data provides a benchmark for cost calculation based on the mold's essential characteristics.

[0111] In some optional embodiments, the method further includes: performing weighted summation on the first cost, the second cost, and the third cost to obtain a comprehensive cost of manufacturing the target mold.

[0112] In this embodiment, the first cost is calculated based on the relationship between time-series process data and cost, reflecting the impact of process changes over time on cost. The second cost is calculated based on the relationship between spatial process data and cost, reflecting the impact of spatial correlations between process parameters and equipment on cost. The third cost is calculated based on the contribution of structured data (such as geometric complexity, process path, and material properties) to cost, taking into account the impact of the mold's inherent properties on cost. Through weighted summation, these three different dimensions of cost information are combined, avoiding the one-sidedness that can result from single-dimensional cost calculations. This comprehensive cost more comprehensively reflects the various cost factors in the mold manufacturing process.

[0113] In some optional embodiments, the calculation formula of the comprehensive cost of the target mold includes the following:

[0114]

[0115] in, The comprehensive cost of the target mold, 、 and Represent the prediction functions based on XGBoost, LSTM and graph convolutional network respectively, For the structured data, is the timing process data, is the spatial process data, 、 and is the gating coefficient, which is dynamically adjusted through the preset attention mechanism.

[0116] In this embodiment, XGBoost is used to model the relationship between structured data and cost. By building multiple decision trees, XGBoost can effectively capture nonlinear relationships and interactions in the data, improving the accuracy and stability of predictions.

[0117] LSTM (Long Short-Term Memory) is a time series model in the spatiotemporal graph neural network (STGCN). It excels at processing time series data and is used to model the primary relationship between time series process data and cost. LSTM effectively captures time-dependent patterns, identifies regularities and trends in time series data, and provides a time series basis for cost forecasting.

[0118] GCN (Graph Convolutional Network), a graph neural network model within the spatiotemporal graph neural network (STGCN), is used to process graph-structured data and model the relationships between process parameters and the secondary relationship between costs. By constructing a correlation graph between process parameters, GCN can capture the mutual influence and synergy between parameters, providing a correlation basis for cost prediction.

[0119] Gating coefficient 、 and Dynamic adjustments through the attention mechanism reflect the importance and contribution of different data sources and models in the current scenario. This dynamic adjustment mechanism enables the model to automatically adjust the weights of each sub-model based on the quality, relevance, and reliability of different data, improving the accuracy and robustness of the overall prediction.

[0120] The attention mechanism dynamically adjusts the gating coefficients (i.e., weights) of XGBoost, LSTM, and GCN based on the characteristics of the input data. This dynamic adjustment enables the model to automatically assign importance to different models in different situations. For example, when structured data has a significant impact on cost, the attention mechanism increases the gating coefficient of XGBoost; when changes in sequential process data significantly affect cost, the gating coefficient of LSTM is increased. This allows the method of this application to adapt to complex and changing manufacturing environments, such as equipment status, material supply, and process adjustments. By dynamically adjusting the gating coefficients through the attention mechanism, the comprehensive cost calculation method can adapt to these complex and changing environments, adjusting model weights in real time, and ensuring that cost predictions remain highly accurate.

[0121] In some optional embodiments, after obtaining the comprehensive cost of manufacturing the target mold, it also includes: based on the comprehensive cost and real-time data of the production line, using a preset reinforcement learning algorithm to calculate a cost control strategy, the cost control strategy includes at least one of process parameter setting, material selection and processing path optimization, and the reward function of the reinforcement learning algorithm includes the degree of cost reduction and process quality; using a preset digital twin model of the production line to verify the cost control strategy, and adjusting the cost control strategy according to the verification results until the verification results meet the preset standards to obtain the final cost control strategy.

[0122] In this embodiment, the production line's digital twin model is a virtual reflection of the actual production line, allowing cost control strategies to be validated without disrupting actual production. By simulating the implementation of strategies within the digital twin model, potential issues and risks, such as equipment conflicts and process mismatches, can be identified in advance, thus avoiding adverse consequences such as increased costs, reduced quality, or production interruptions in actual production.

[0123] The preset reinforcement learning algorithm is used to generate the optimal cost control strategy based on the cost prediction results. This embodiment uses the Deep Q Network (DQN) as the reinforcement learning framework to achieve dynamic optimization and adjustment of costs. The implementation of the DQN network is shown in the code:

[0124] ```Python

[0125] class DQNAgent:

[0126] def __init__(self):

[0127] self.memory = ReplayBuffer(10000)

[0128] self.q_net = QNetwork(256)

[0129] def choose_action(self, state):

[0130] state = torch.FloatTensor(state)

[0131] return self.q_net(state).argmax()

[0132] ```

[0133] In this implementation, the DQNAgent class contains an experience replay buffer and a Q network. The experience replay buffer is used to store historical experience, and the Q network is used to learn the mapping between states and actions. The choose_action method uses the Q network to select the optimal action based on the current state.

[0134] In the context of mold cost control, the reinforcement learning environment state consists of the current cost situation, process parameters, and equipment status. Actions include various cost control measures, such as adjusting process parameters, changing materials, and optimizing processing paths. The reward function is defined as a comprehensive indicator of cost reduction and process quality, taking into account both cost control effectiveness and the impact of process quality.

[0135] Reinforcement learning continuously optimizes its algorithmic approach through trial and error. In mold manufacturing, reinforcement learning algorithms can continuously adjust processing parameters to find the optimal manufacturing solution. For example, by adjusting cutting speed and feed rate, the optimal balance between processing efficiency and cost can be achieved. Reinforcement learning (RL) is a machine learning technique that enables robots to make intelligent decisions by learning from experience. By receiving programmed rewards or penalties, the AI model driving the robot continuously improves through a process of trial and error.

[0136] Through a reinforcement learning optimizer, the system dynamically adjusts cost control strategies based on real-time data and historical experience, minimizing costs and optimizing process quality. This data-driven decision-making approach is more scientific and accurate than traditional manual judgment, effectively improving the effectiveness and efficiency of cost control.

[0137] In some optional embodiments, the extracting feature data related to the target mold manufacturing cost from the multimodal data includes: extracting initial feature data related to cost calculation based on the multimodal data; and performing feature dimension compression on the initial feature data to obtain the final feature data related to cost calculation.

[0138] This example summarizes that high-dimensional data is difficult to visualize directly. However, by reducing the data to a two- or three-dimensional space through dimensional compression, the relationships and distribution of the data can be more intuitively displayed. For example, visualization tools such as scatter plots and line graphs can be used to display the relationship between compressed feature data and costs, helping to better understand the factors affecting costs.

[0139] Specifically, during data processing, we use a combination of PCA (Principal Component Analysis) and Autoencoder to achieve feature dimension compression. This not only reduces computational complexity and improves system efficiency, but also reduces data redundancy and improves the quality and effectiveness of feature representation.

[0140] Example 2:

[0141] Based on the techniques of the above embodiments, this embodiment provides an application example. This application provides a mold cost calculation system. The system applies the cost calculation method described in the above embodiments.

[0142] Specifically, the overall architecture of the system is as follows: Figure 2 As shown, in this architecture,

[0143] The physical layer collects various sensor data, equipment status data, and environmental parameter data through the OPC UA protocol to achieve real-time perception of the entire manufacturing process;

[0144] Edge computing nodes are responsible for preliminary data processing and analysis, alleviating back-end computing pressure;

[0145] The Kafka streaming platform serves as the core data bus of the system, responsible for efficient data transmission and processing, and supporting the system's real-time response capabilities;

[0146] The feature engineering pipeline cleans, transforms, and extracts features from raw data to provide standardized, high-quality input data for subsequent model calculations;

[0147] The hybrid prediction model combines the advantages of multiple machine learning algorithms to achieve accurate prediction of mold costs;

[0148] The dynamic adjustment engine generates and executes cost control strategies based on forecast results and real-time data;

[0149] Digital twin verification simulates and verifies the control strategy to ensure its effectiveness and safety;

[0150] Finally, the process optimization recommendations present the optimization results to users in an understandable form to support decision making.

[0151] Based on the above architecture, the system processes 12 types of heterogeneous data sources, including design data, process data, equipment data, environmental data, quality data, etc. These data come from diverse sources and formats and are processed and integrated through a unified data governance framework. First, for structured data, technologies such as data cleaning, standardization, and feature engineering are used to extract meaningful features and indicators. For unstructured data, such as design drawings, processing procedures, and inspection reports, deep learning models are used for semantic understanding and information extraction. For time series data, time series analysis and prediction techniques are used to mine time-dependent patterns. For spatial data, such as three-dimensional mold models and processing paths, spatial analysis techniques are used to extract geometric features and topological relationships.

[0152] Through multimodal data fusion, the system achieves comprehensive perception and understanding of the entire mold manufacturing process, providing rich data support for subsequent analysis and decision-making. During data processing, a combination of PCA (Principal Component Analysis) and autoencoders is used to achieve feature dimensionality compression, achieving a compression rate of 85%. This not only reduces computational complexity and improves system efficiency, but also reduces data redundancy and enhances the quality and effectiveness of feature representation.

[0153] Feature engineering technology is the core of machine learning and has a decisive impact on model performance. This embodiment designs a dynamic feature engineering method for multi-source heterogeneous data in the mold manufacturing process to maximize the mining of data value. The system defines three key features: geometric complexity, process path, and material properties, as shown in the following table:

[0154]

[0155] Geometric complexity: Features extracted through 3D point cloud curvature analysis reflect the complexity of the mold shape and its impact on machining difficulty and cost. Research has shown that mold geometric complexity directly affects machining process complexity, which in turn affects machining costs. Curvature analysis can quantify the degree of curvature of the mold surface, identify areas of high complexity, and provide a geometric basis for cost estimation.

[0156] Process path: Features are extracted through G-code semantic parsing, reflecting the planning and execution of the machining path. G-code is the core instruction in CNC machining and contains a wealth of process information. Through semantic parsing, the system can extract key parameters such as path length, feed rate, and cutting depth from the code, providing a process basis for cost calculation. Optimizing the process path is crucial for reducing machining costs, especially in high-variety, low-volume production models.

[0157] Material Characteristics: Characteristics extracted through dynamic hardness mapping reflect changes in the material's physical properties during processing. Material characteristics directly impact processing difficulty and cost, and different materials require different processing techniques and parameter settings. Through dynamic hardness mapping analysis, the system can identify the material's hardness distribution and variation patterns, providing a material basis for optimizing process parameters. In mold manufacturing, material selection and performance analysis are critical links, directly affecting mold quality and cost.

[0158] These three types of features together constitute the feature space for mold cost calculation, covering the main influencing factors in the mold manufacturing process. Through dynamic feature engineering, the system can dynamically adjust feature weights and calculation methods based on real-time data and historical experience to adapt to different scenarios and changing needs. This dynamic nature makes the system highly adaptable and robust, enabling it to cope with various changes and challenges in the mold manufacturing process.

[0159] The Spatiotemporal Graph Network (STGCN) model is the system's core algorithm module, used to model and analyze the spatiotemporal relationships between process parameters and cost factors. The STGCN model combines the strengths of graph neural networks and time series models to effectively capture spatiotemporal dependencies in data.

[0160] The structure of the STGCN model is shown in the code:

[0161] ```Python

[0162] class STGCN(nn.Module):

[0163] def __init__(self):

[0164] self.gcn = GraphConv(in_dim=128, out_dim=64)

[0165] self.tcn = TemporalConv(64, 32)

[0166] def forward(self, graph_data):

[0167] spatial_feat = self.gcn(graph_data)

[0168] temporal_feat = self.tcn(spatial_feat)

[0169] return temporal_feat

[0170] ```

[0171] In this model, the GraphConv layer captures spatial relationships, such as the mutual influence between different process parameters and the collaborative relationships between different equipment. The TemporalConv layer captures temporal patterns, such as the trends in process parameters over time and the dynamic evolution of cost factors. By combining these two layers, STGCN can comprehensively and deeply analyze the complex relationship between process and cost.

[0172] Spatiotemporal graph neural networks (STGNNs) have garnered widespread attention in recent years. By integrating graph neural networks with various temporal learning methods, they are able to extract complex spatiotemporal dependencies. In urban computing forecasting, the STGNN framework has demonstrated strong performance, effectively processing spatiotemporal data and delivering accurate predictions. In multivariate time series forecasting, models based on dynamic adaptive spatiotemporal graphs have demonstrated superior performance. These studies demonstrate the significant advantages of spatiotemporal graph neural networks in processing data with spatiotemporal characteristics, providing a theoretical foundation for their application in mold cost control.

[0173] Using the STGCN model, the system can construct a process-cost correlation matrix to quantitatively analyze the degree and direction of the impact of process parameters on costs. This matrix provides a clear direction and basis for cost optimization, guiding companies to implement targeted process improvements and cost control.

[0174] The hybrid prediction model is the core algorithm module of the system, used to accurately predict mold costs. The model combines the advantages of multiple machine learning algorithms to effectively model complex nonlinear relationships. The mathematical expression of the hybrid prediction model is:

[0175]

[0176] in, The comprehensive cost of the target mold, 、 and Represent the prediction functions based on XGBoost, LSTM and graph convolutional network respectively, For the structured data, is the timing process data, is the spatial process data, 、 and is the gating coefficient, which is dynamically adjusted through the preset attention mechanism.

[0177] XGBoost is an ensemble learning algorithm known for its high performance and interpretability. In this example, XGBoost is used to model the relationship between structured data (such as material properties and equipment parameters) and costs. By building multiple decision trees, XGBoost effectively captures nonlinear relationships and interactions in the data, improving the accuracy and stability of predictions.

[0178] LSTM (Long Short-Term Memory) is a recurrent neural network that excels at processing time series data. In this example, LSTM is used to model the relationship between time series data (such as parameter changes and environmental changes during the processing process) and costs. LSTM can effectively capture time-dependent patterns, identify regularities and trends in time series data, and provide a temporal basis for cost forecasting.

[0179] A graph convolutional network (GCN) is a type of graph neural network used to process graph-structured data. In this example, GCN is used to model the relationship between process parameters and costs. By constructing a correlation graph between process parameters, GCN can capture the mutual influence and synergy between parameters, providing a correlation basis for cost prediction.

[0180] The gating coefficients α, β, and γ are dynamically adjusted through the attention mechanism, reflecting the importance and contribution of different data sources and models in the current scenario. This dynamic adjustment mechanism enables the model to automatically adjust the weights of each sub-model based on the quality, relevance, and reliability of different data, improving the accuracy and robustness of the overall prediction.

[0181] The advantage of hybrid forecasting models lies in their ability to combine the characteristics of different models, complementing their respective deficiencies and improving the comprehensiveness and accuracy of forecasts. XGBoost provides interpretability and stability, LSTM captures time series features, and GCN models correlations. These three combine to form a comprehensive and accurate forecasting framework. This hybrid approach offers significant advantages in complex system modeling and can effectively address the various challenges in mold cost forecasting.

[0182] The reinforcement learning optimizer is the system's core control module, used to generate the optimal cost control strategy based on cost prediction results. This embodiment uses the Deep Q Network (DQN) as the reinforcement learning framework to achieve dynamic cost optimization and adjustment.

[0183] The implementation of the DQN network is shown in the code:

[0184] ```Python

[0185] class DQNAgent:

[0186] def __init__(self):

[0187] self.memory = ReplayBuffer(10000)

[0188] self.q_net = QNetwork(256)

[0189] def choose_action(self, state):

[0190] state = torch.FloatTensor(state)

[0191] return self.q_net(state).argmax()

[0192] ```

[0193] In this implementation, the DQNAgent class contains an experience replay buffer and a Q network. The experience replay buffer is used to store historical experience, and the Q network is used to learn the mapping between states and actions. The choose_action method uses the Q network to select the optimal action based on the current state.

[0194] In the context of mold cost control, the reinforcement learning environment state consists of the current cost situation, process parameters, and equipment status. Actions include various cost control measures, such as adjusting process parameters, changing materials, and optimizing processing paths. The reward function is defined as a comprehensive indicator of cost reduction and process quality, taking into account both cost control effectiveness and the impact of process quality.

[0195] Reinforcement learning continuously optimizes its algorithmic approach through trial and error. In mold manufacturing, reinforcement learning algorithms can continuously adjust processing parameters to find the optimal manufacturing solution. For example, by adjusting cutting speed and feed rate, the optimal balance between processing efficiency and cost can be achieved. Reinforcement learning (RL) is a machine learning technique that enables robots to make intelligent decisions by learning from experience. By receiving programmed rewards or penalties, the AI model driving the robot continuously improves through a process of trial and error.

[0196] Through a reinforcement learning optimizer, the system dynamically adjusts cost control strategies based on real-time data and historical experience, minimizing costs and optimizing process quality. This data-driven decision-making approach is more scientific and accurate than traditional manual judgment, effectively improving the effectiveness and efficiency of cost control.

[0197] In order to verify the effectiveness of the system in this embodiment, a practical application test was conducted in a large mold company. The experimental environment configuration is shown in Table 2 below:

[0198]

[0199] Table 2 (Configuration Table)

[0200] The hardware platform used in the experiment was an NVIDIA DGX A100 cluster, which provided powerful computing power to support deep learning and the efficient execution of complex algorithms. The data volume reached 15TB of CAD data and 240 million sensor data points, covering the entire mold design, processing, assembly, and inspection process, providing rich data support for system training and testing.

[0201] The SAP MES (Manufacturing Execution System) and SolidWorks PLM (Product Lifecycle Management) systems were selected as the comparison benchmarks. These are widely used management software in the mold industry and have high market recognition and application value. By comparing these mature systems, we can objectively evaluate the advantages and innovations of the method proposed in this embodiment.

[0202] To verify the performance advantages of the system, this embodiment compares the traditional system and the present system in terms of cost prediction accuracy, abnormal response time, and dynamic adjustment success rate. The results are shown in Table 3:

[0203]

[0204] Table 3 (Performance comparison)

[0205] Cost forecast accuracy: The mean absolute error (MAE) of the traditional system was 87,000 yuan, while the MAE of this system was only 31,000 yuan, a 64.4% reduction. This demonstrates the significant advantages of this system in cost forecasting, enabling more accurate mold cost estimates and providing a reliable basis for cost control. Cost forecast accuracy is crucial for business decision-making, and accurate forecasts can guide enterprises in more rational resource allocation and cost management.

[0206] Abnormal response time: Traditional systems have an abnormality detection delay of 4 hours, while this system's abnormality detection delay is only 8.7 seconds, a 99.4% reduction in response time. This demonstrates the system's significant advantages in real-time monitoring and rapid response, enabling timely detection and resolution of cost anomalies, preventing escalation and increased losses. In a rapidly changing market environment, timely response capabilities are crucial to a company's survival and development.

[0207] Dynamic Adjustment Success Rate: While the traditional system had a dynamic adjustment success rate of 68%, this system achieved a 92.3% success rate, a 35.7% improvement. This demonstrates that this system offers significant advantages in generating and executing cost control strategies, enabling more effective cost control objectives. Dynamic adjustment capability is the core competitiveness of intelligent cost control systems, determining their ability to respond to change and challenges.

[0208] These comparison results fully demonstrate the effectiveness and superiority of the method proposed in this example, verifying the practical application value of the system. Through the application of artificial intelligence and machine learning technologies, the system achieves intelligent, dynamic, and refined cost control, significantly improving the company's cost management level and competitiveness.

[0209] In order to more intuitively demonstrate the optimization effect of the system, this embodiment conducts a comparative analysis of the various components of the mold cost. The results are as follows: Figure 3 As shown in the figure, the system has achieved significant optimization in various cost components:

[0210] 1. Material Cost: The traditional method used 42 units of material cost, which was reduced to 35 units after optimization, a 16.7% reduction. This demonstrates that the system can effectively reduce material costs by optimizing material selection and reducing material waste. Material cost is a major cost component in mold manufacturing, accounting for approximately 40% of the total cost. Reducing material cost is crucial for overall cost control.

[0211] 2. Processing Cost: The traditional processing cost was 33 units, but after optimization, it was reduced to 28 units, a 15.2% reduction. This demonstrates that the system can effectively reduce processing costs by optimizing the processing technology and improving processing efficiency. Processing cost is the second largest cost component in mold manufacturing, accounting for approximately 30% of the total cost. Reducing processing cost is crucial for improving a company's competitiveness.

[0212] 3. Energy Cost: The energy cost under the traditional method was 15 units, but after optimization, it dropped to 9 units, a 40% reduction. This demonstrates that the system can effectively reduce energy costs by optimizing equipment operation and reducing energy waste. Although energy costs account for a relatively small proportion of total costs, reducing energy consumption not only saves money but also reduces environmental pollution and enhances the company's social responsibility image.

[0213] These optimization results demonstrate that the system of this embodiment not only reduces overall costs but also precisely optimizes different cost components, improving the level of refinement in cost control. This comprehensive and precise cost control capability is difficult to achieve with traditional methods.

[0214] In summary, this embodiment of the mold cost calculation system is specifically manifested in three aspects: methodological innovation, technological innovation, and application innovation:

[0215] 1. Methodological innovation: process feature spatiotemporal graph modeling theory

[0216] This paper proposes a new methodological framework—the theory of process feature spatiotemporal graph modeling—which provides new ideas and tools for mold cost control. The core of this methodological framework is the spatiotemporal graph neural network (STGCN), which integrates graph neural networks with various temporal learning methods to extract complex spatiotemporal dependencies.

[0217] The STGCN model simultaneously considers the spatial correlation and temporal evolution of process parameters, comprehensively capturing the complex relationship between process and cost. In the spatial dimension, STGCN uses graph convolution to capture the mutual influence and correlation between different process parameters. In the temporal dimension, STGCN uses temporal convolution to capture the dynamic changes in process parameters and cost factors. By combining space and time, STGCN enables a more comprehensive and in-depth analysis of the relationship between process and cost, providing more accurate predictions and more effective strategies for cost control.

[0218] The innovation of the process feature spatiotemporal graph modeling theory lies in its focus not only on the impact of individual process parameters on cost, but also on the interactions and synergies between process parameters, as well as how these influences and effects vary over time and space. This comprehensive, dynamic modeling approach more accurately reflects the complexity and variability of the mold manufacturing process, providing a more scientific and precise basis for cost control.

[0219] 2. Technological innovation: dynamic adjustment engine driven by streaming computing

[0220] This embodiment provides a new technical architecture—a dynamic adjustment engine driven by streaming computing—that provides powerful technical support for mold cost control. The core of this technical architecture is streaming computing technology, which can analyze large amounts of flowing data in real time during a continuously changing process, capture valuable information, and send the results to the next computing node.

[0221] The dynamic adjustment engine, driven by streaming computing, monitors various manufacturing process data in real time, such as equipment status, processing parameters, and environmental conditions, to promptly identify cost anomalies and automatically generate and execute adjustment measures based on pre-set rules or algorithmic models. This real-time monitoring and rapid response capability enables the system to promptly identify and address cost issues, preventing them from escalating and increasing losses.

[0222] The innovation of the dynamic adjustment engine driven by streaming computing lies in its ability to break the cyclical and lag-based nature of traditional cost control, achieving real-time and forward-looking cost control. System response time has been reduced from the traditional 4-6 hours to 8.7 seconds. This real-time response capability enables the system to promptly detect and address cost anomalies, preventing them from escalating and increasing losses. In a rapidly changing market environment, timely response capabilities are crucial to a company's survival and development.

[0223] 3. Application innovation: closed-loop control system verified by digital twins

[0224] This example establishes a new application model—a closed-loop control system for digital twin verification—providing a novel application scenario for mold cost control. The core of this application model is digital twin technology, which creates virtual replicas of physical objects or systems, enabling simulation, testing, and optimization in a virtual environment.

[0225] The closed-loop control system validated by digital twins maps the actual manufacturing process into a virtual environment, creating a digital twin model. This model not only reflects the actual process status and changes in real time but can also be used to simulate and test various cost control strategies, predicting their effectiveness and impact. Through digital twin validation, the system can evaluate the effectiveness and safety of various cost control strategies before implementation, reducing implementation risks and costs.

[0226] The innovation of the closed-loop control system validated by digital twins lies in its closed-loop management of cost control, forming a complete process of "prediction-decision-execution-evaluation." Through digital twin validation, the system can continuously learn and optimize, improving the effectiveness and efficiency of cost control. Linking the digital twin model with the production system enables synchronized operation and real-time monitoring, enabling virtual simulation of the product design and manufacturing process, thereby improving product quality, production efficiency, and product quality stability.

[0227] The mold cost calculation system of this embodiment achieves three major breakthroughs in the field of mold cost control:

[0228] First, the cost forecast accuracy surpasses the industry benchmark (<5% error). While traditional methods have a cost forecast error rate as high as 15.6%, this system reduces the error rate to 3.1%, a reduction of 80%. This demonstrates the system's significant advantages in cost forecasting, enabling more accurate mold cost estimates and providing a reliable basis for cost control. Cost forecast accuracy is crucial for business decision-making, and accurate forecasts can guide more rational resource allocation and cost management.

[0229] Secondly, dynamic response speeds reach sub-second levels. While traditional systems experience anomaly detection delays of four hours, this system achieves a mere 8.7 seconds, a 99.4% reduction in response time. This demonstrates the system's significant advantages in real-time monitoring and rapid response, enabling timely detection and resolution of cost anomalies, preventing them from escalating and increasing losses. In a rapidly changing market environment, timely response is crucial for business survival and growth.

[0230] Third, an explainable process optimization knowledge base has been constructed. The system not only enables cost prediction and control but also provides explainable optimization recommendations and knowledge, helping users understand and apply these optimization measures. This explainability makes the system more than just a black-box tool; it becomes a knowledge platform that promotes user learning and growth.

[0231] These breakthroughs fully demonstrate the value of applying artificial intelligence and machine learning technologies to mold cost control and showcase the advantages and potential of data-driven decision-making. By applying these advanced technologies to mold cost control, we have achieved a shift from experience-driven to data-driven, from passive response to proactive prevention, and from extensive management to refined management.

[0232] Example 3:

[0233] Another embodiment of the present application relates to a mold cost calculation device. The implementation details of the mold cost calculation device of this embodiment are described in detail below. The following content is only for the convenience of understanding the implementation details and is not necessary for the implementation of this solution. The schematic diagram of the mold cost calculation device of this embodiment can be as follows: Figure 4 As shown, it includes a data acquisition module 410 , a feature data extraction module 420 , a first cost calculation module 430 , and a second cost calculation module 440 .

[0234] The data acquisition module 410 is used to acquire multimodal data related to target mold manufacturing;

[0235] a feature data extraction module 420 for extracting feature data related to the target mold manufacturing cost from the multimodal data;

[0236] a first cost calculation module 430 configured to extract time series process data from the feature data based on a preset spatiotemporal graph network model, analyze a first relationship between the time series process data and the cost, and calculate a first cost of the target mold based on the first relationship; and / or

[0237] The second cost calculation module 440 is used to extract spatial process data from the feature data based on the spatiotemporal graph network model, and analyze the second relationship between the spatial process data and the cost, and calculate the second cost of the target mold according to the second relationship, wherein the spatial process data includes interrelated process parameters and / or interrelated equipment.

[0238] It is worth mentioning that all modules involved in this embodiment are logical modules. In actual applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. In addition, to highlight the innovation of this application, this embodiment does not include units that are not closely related to solving the technical problem proposed by this application. However, this does not mean that other units do not exist in this embodiment.

[0239] In some optional embodiments, the mold cost calculation device further includes:

[0240] an association matrix construction module, configured to construct a process-cost association matrix based on the first relationship and the second relationship, wherein the process-cost association matrix is used to quantitatively analyze the degree and direction of influence of characteristic data on cost;

[0241] The associated cost calculation module is used to calculate the first cost and the second cost according to the process-cost association matrix.

[0242] In some optional embodiments, the incidence matrix building module includes:

[0243] a space-time graph construction unit, configured to abstract the characteristic data and cost elements into nodes in a space-time graph structure using the space-time graph network model, wherein the cost elements include at least one of raw material procurement cost, processing cost, and processing loss cost;

[0244] A first relationship building unit is configured to analyze the historical sequence data of each node through a time series model in the spatiotemporal graph network model to obtain the first relationship between the feature data and the cost element;

[0245] A second relationship construction unit is configured to aggregate information of neighbor nodes through a graph neural network model in the spatiotemporal graph network model to obtain the second relationship between the feature data and the cost element;

[0246] The correlation matrix construction unit is used to construct the process-cost correlation matrix according to the first relationship and the second relationship.

[0247] In some optional embodiments, the device further comprises:

[0248] a structured data extraction module, configured to extract structured data from the feature data, wherein the type of the structured data includes at least one of geometric complexity, process path, and material properties;

[0249] The third cost calculation module is configured to calculate the third cost of the target mold according to the contribution of the structured data to the cost.

[0250] In some optional embodiments, the structured data extraction module includes at least one of the following:

[0251] A geometric complexity extraction unit, configured to extract the geometric complexity by performing 3D point cloud curvature analysis; and

[0252] a process path extraction unit, configured to extract relevant parameters of the process path based on a numerical control machining instruction library, wherein the relevant parameters include at least one of a path length, a feed rate, and a cutting depth; and

[0253] The material property extraction unit is used to obtain the material property through dynamic hardness map analysis, wherein the dynamic hardness map is used to simulate the change of material hardness with position, temperature, time or strain rate under actual processing conditions.

[0254] In some optional embodiments, the device further comprises:

[0255] The comprehensive cost calculation module is used to perform weighted summation on the first cost, the second cost and the third cost to obtain the comprehensive cost of manufacturing the target mold.

[0256] In some optional embodiments, the method further includes:

[0257] a cost control strategy generation module, configured to calculate a cost control strategy based on the comprehensive cost and real-time production line data using a preset reinforcement learning algorithm, wherein the cost control strategy includes at least one of process parameter setting, material selection, and processing path optimization, and the reward function of the reinforcement learning algorithm includes the degree of cost reduction and process quality;

[0258] The control strategy verification module is used to verify the cost control strategy using the digital twin model of the preset production line, and adjust the cost control strategy according to the verification results until the verification results meet the preset standards to obtain the final cost control strategy.

[0259] In some optional embodiments, the feature data extraction module includes:

[0260] an initial feature data extraction unit, configured to extract initial feature data related to cost calculation based on the multimodal data;

[0261] The feature dimension compression unit is used to perform feature dimension compression on the initial feature data to obtain the feature data finally related to cost calculation.

[0262] Example 4:

[0263] Another embodiment of the present application relates to an electronic device, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to execute the mold cost calculation method in the above-mentioned embodiments.

[0264] The memory and processor are connected using a bus, which can include any number of interconnected buses and bridges. The bus connects various circuits of one or more processors and memories. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits. These are all well known in the art and are therefore not described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single component or multiple components, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor is transmitted over a wireless medium via an antenna. Furthermore, the antenna receives data and transmits it to the processor.

[0265] The processor is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. Memory can be used to store data used by the processor when performing operations.

[0266] Embodiment 5:

[0267] Another embodiment of the present application relates to a computer-readable storage medium storing a computer program, which implements the above method embodiment when executed by a processor.

[0268] That is, those skilled in the art will understand that all or part of the steps in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a program. The program is stored in a storage medium and includes a number of instructions for causing a device (which may be a single-chip microcomputer, chip, etc.) or a processor to execute all or part of the steps in the methods described in the various embodiments of this application. The aforementioned storage medium includes various media that can store program code, such as a USB flash drive, a mobile hard drive, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0269] Those skilled in the art will appreciate that the above embodiments are specific embodiments for implementing the present application, and that in actual applications, various changes may be made thereto in form and detail without departing from the spirit and scope of the present application.

Claims

1. A mold cost calculation method, characterized in that: include: Acquire multimodal data related to target mold manufacturing; extracting feature data related to the target mold manufacturing cost from the multimodal data; Extracting time-series process data from the feature data based on a preset spatiotemporal graph network model, analyzing a first relationship between the time-series process data and the cost, and calculating a first cost of the target mold based on the first relationship; Extracting spatial process data from the feature data based on the spatiotemporal graph network model, analyzing a second relationship between the spatial process data and cost, and calculating a second cost of the target mold based on the second relationship, wherein the spatial process data includes interrelated process parameters and / or interrelated equipment; Calculating a first cost of the target mold according to the first relationship, and calculating a second cost of the target mold according to the second relationship, comprising: Constructing a process-cost association matrix based on the first relationship and the second relationship, wherein the process-cost association matrix is used to quantitatively analyze the degree and direction of influence of characteristic data on cost; Calculating the first cost and the second cost according to the process-cost association matrix; The constructing of a process-cost association matrix according to the first relationship and the second relationship includes: Abstracting the characteristic data and cost elements into nodes in a space-time graph structure using the space-time graph network model, wherein the cost elements include at least one of raw material procurement cost, processing cost, and processing loss cost; Analyzing the historical sequence data of each node through the time series model in the space-time graph network model to obtain the first relationship between the feature data and the cost element; aggregating information of neighbor nodes through a graph neural network model in the spatiotemporal graph network model to obtain the second relationship between the feature data and the cost element; Constructing the process-cost association matrix according to the first relationship and the second relationship; The method further comprises: Extracting structured data from the feature data, wherein the type of the structured data includes at least one of geometric complexity, process path, and material properties; Calculating a third cost of the target mold according to the contribution of the structured data to the cost; The method further comprises: A weighted sum is taken for the first cost, the second cost, and the third cost to obtain a comprehensive cost for manufacturing the target mold.

2. The mold cost calculation method according to claim 1, characterized in that: The extracting structured data from the feature data includes at least one of the following: The geometric complexity is extracted by 3D point cloud curvature analysis; or Extracting relevant parameters under the process path based on a numerical control machining instruction library, wherein the relevant parameters include at least one of path length, feed speed and cutting depth; or The material properties are obtained by dynamic hardness map analysis, wherein the dynamic hardness map is used to simulate the change of material hardness with position, temperature, time or strain rate under actual processing conditions.

3. The mold cost calculation method according to claim 1, characterized in that: The calculation formula of the comprehensive cost of the target mold includes the following: in, The comprehensive cost of the target mold, 、 and Represent the prediction functions based on XGBoost, LSTM and graph convolutional network respectively, For the structured data, is the timing process data, is the spatial process data, 、 and is the gating coefficient, which is dynamically adjusted through the preset attention mechanism.

4. The mold cost calculation method according to claim 1, characterized in that: After obtaining the comprehensive cost of manufacturing the target mold, the method further includes: Based on the comprehensive cost and real-time production line data, a cost control strategy is calculated using a preset reinforcement learning algorithm, wherein the cost control strategy includes at least one of process parameter setting, material selection, and processing path optimization, and the reward function of the reinforcement learning algorithm includes the degree of cost reduction and process quality; The cost control strategy is verified using a digital twin model of a preset production line, and the cost control strategy is adjusted according to the verification results until the verification results meet the preset standards to obtain the final cost control strategy.

5. The mold cost calculation method according to any one of claims 1 to 4, characterized in that: The extracting feature data related to the target mold manufacturing cost from the multimodal data includes: Extracting initial feature data related to cost calculation based on the multimodal data; The initial feature data is subjected to feature dimension compression to obtain the feature data finally related to cost calculation.

6. A mold cost calculation device, characterized in that: include: A data acquisition module, used to acquire multimodal data related to target mold manufacturing; a feature data extraction module, configured to extract feature data related to the target mold manufacturing cost from the multimodal data; a first cost calculation module, configured to extract time series process data from the feature data based on a preset spatiotemporal graph network model, analyze a first relationship between the time series process data and the cost, and calculate a first cost of the target mold based on the first relationship; a second cost calculation module, configured to extract spatial process data from the feature data based on the spatiotemporal graph network model, analyze a second relationship between the spatial process data and cost, and calculate a second cost of the target mold according to the second relationship, wherein the spatial process data includes interrelated process parameters and / or interrelated equipment; The mold cost calculation device also includes: an association matrix construction module, configured to construct a process-cost association matrix based on the first relationship and the second relationship, wherein the process-cost association matrix is used to quantitatively analyze the degree and direction of influence of characteristic data on cost; an associated cost calculation module, configured to calculate the first cost and the second cost according to the process-cost association matrix; The correlation matrix building module includes: a space-time graph construction unit, configured to abstract the characteristic data and cost elements into nodes in a space-time graph structure using the space-time graph network model, wherein the cost elements include at least one of raw material procurement cost, processing cost, and processing loss cost; A first relationship building unit is configured to analyze the historical sequence data of each node through a time series model in the spatiotemporal graph network model to obtain the first relationship between the feature data and the cost element; A second relationship construction unit is configured to aggregate information of neighbor nodes through a graph neural network model in the spatiotemporal graph network model to obtain the second relationship between the feature data and the cost element; A correlation matrix construction unit, configured to construct the process-cost correlation matrix according to the first relationship and the second relationship; The device further comprises: a structured data extraction module, configured to extract structured data from the feature data, wherein the type of the structured data includes at least one of geometric complexity, process path, and material properties; a third cost calculation module, configured to calculate a third cost of the target mold according to the contribution of the structured data to the cost; The device further comprises: The comprehensive cost calculation module is used to perform weighted summation on the first cost, the second cost and the third cost to obtain the comprehensive cost of manufacturing the target mold.

7. An electronic device, characterized in that: include: at least one processor; as well as, a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the mold cost calculation method according to any one of claims 1 to 5.

8. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the mold cost calculation method according to any one of claims 1 to 5 is implemented.

Images