Intelligent prediction method for die stamping manufacturing cost

Abstract

The invention relates to the technical field of die stamping manufacturing, and discloses an intelligent prediction method for die stamping manufacturing cost. The method comprises the following steps: firstly, acquiring a multi-dimensional manufacturing data set consisting of material characteristic parameters, process processing parameters and equipment operation parameters; then key process characteristics are extracted based on material characteristic parameters, a process influence topological graph is generated according to process processing parameters, and equipment operation parameters are processed to obtain an equipment state evolution trend; and then inputting the data into a multi-factor evaluation model to generate a cost prediction demand vector, finally constructing an adaptive prediction strategy graph through a dynamic programming algorithm, and outputting a manufacturing cost prediction scheme. According to the method, the multi-dimensional data is comprehensively considered, the dynamic change of the data can be effectively processed, the accuracy and reliability of predicting the die stamping manufacturing cost are improved, and an enterprise can reasonably control the cost, optimize the production decision and improve the market competitiveness.

Description

Technical Field

[0001] The present invention relates to the technical field of die stamping manufacturing, in particular to an intelligent prediction method for die stamping manufacturing costs. Background Art

[0002] In the die stamping manufacturing industry, cost control has always been one of the key factors for companies to maintain their competitiveness. With the increasingly fierce market competition and the continuous development of the manufacturing industry, accurately predicting die stamping manufacturing costs has become increasingly important.

[0003] Traditional die stamping manufacturing cost forecasting methods mostly rely on manual experience and simple calculation models. Manual judgment is limited by individual knowledge and practical experience, resulting in significant discrepancies between different personnel, making it difficult to ensure accurate and stable forecasts. For example, when assessing material costs, they rely solely on price estimates based on historically used similar materials, ignoring factors such as market price fluctuations and differences in material properties. When calculating processing costs, they simply multiply conventional processing time and equipment rates, failing to consider the impact of process details on processing efficiency.

[0004] Simple calculation models usually only consider a single or a few cost-influencing factors and cannot fully reflect the complexity of the die stamping manufacturing process. For example, some models only focus on material costs and basic processing hours, but ignore the significant impact of die wear, changes in equipment operating status, etc. on costs. During the stamping process, the degree of wear of the die increases with the number of uses, which not only affects product quality, but may also lead to reduced processing efficiency and increase the cost of repairing or replacing the die. When the equipment is running, different operating parameters will lead to different energy consumption, and the frequency of equipment failure is also closely related to the operating status. These will directly or indirectly increase manufacturing costs, but traditional models find it difficult to take these factors into consideration.

[0005] Furthermore, the die stamping manufacturing process involves multiple types of data, such as material properties, process parameters, and equipment operating parameters. These data are interrelated and dynamically changing. Existing forecasting methods are unable to effectively handle multi-dimensional, dynamically changing data, making it difficult to accurately reflect cost trends in real time. In actual production, material properties vary from batch to batch, process parameters may need to be adjusted based on product requirements, and equipment operating conditions are subject to change at any time. Traditional methods are unable to cope with these complex situations and are unable to provide companies with reliable cost forecasts in a timely manner, which in turn affects their production decisions, cost control, and profit generation.

[0006] With the development of intelligent manufacturing technology, data collection and processing capabilities are constantly improving, providing new opportunities for solving the problem of mold stamping manufacturing cost prediction. However, the application of related technologies in the mold stamping manufacturing field is not yet mature enough, and its advantages have not been fully utilized to achieve accurate cost prediction. Therefore, it is urgent to develop an intelligent mold stamping manufacturing cost prediction method and system that can comprehensively consider multi-dimensional manufacturing data, effectively handle dynamic data changes, and achieve high prediction accuracy. This is of great significance for improving the market competitiveness and economic benefits of mold stamping manufacturing companies. Summary of the Invention

[0007] The purpose of the present invention is to provide an intelligent prediction method for mold stamping manufacturing costs to solve the problems raised in the above background technology.

[0008] To achieve the above-mentioned object, the present invention provides the following technical solution: a method for intelligently predicting mold stamping manufacturing costs, the method comprising: Acquire a multi-dimensional manufacturing data set; the multi-dimensional manufacturing data includes material characteristic parameters, process parameters, and equipment operating parameters; the material characteristic parameters include material ductility coefficient and hardness distribution gradient, and the process parameters include stamping speed adjustment value and die wear index; Based on the material characteristic parameters, extracting key process characteristics through process parameter analysis, wherein the key process characteristics include material deformation threshold, stress concentration factor and energy consumption fluctuation range; According to the process parameters, a process influence topology map is generated by a characteristic association network construction algorithm, wherein the influence topology map includes parameter interaction paths and node influence weights; Processing the equipment operating parameters in time series segments to generate equipment status evolution trends; Inputting the key process characteristics, impact topology and equipment status evolution trend into a multi-factor evaluation model to generate a cost prediction demand vector; Based on the predicted demand vector, an adaptive prediction strategy graph is constructed through a dynamic programming algorithm to output a manufacturing cost prediction solution; the nodes of the adaptive prediction strategy graph represent prediction analysis modules, and the edges represent analysis order and complexity weight.

[0009] Preferably, the extracting key process features through process parameter analysis includes: Eliminating abnormal values from the material characteristic parameters to generate standardized material data; Based on the preset benchmark process parameter library, the basic deformation categories are identified through the gradient matching algorithm and the deformation stages are divided; The frequency domain decomposition algorithm is used to extract the energy distribution characteristics of the deformation stage, and the deformation deviation is generated by combining the reference threshold in the benchmark process parameter library; The deformation deviation, stress concentration factor and energy consumption fluctuation range are encoded as key process characteristics.

[0010] Preferably, generating a process impact topology map by constructing an algorithm using a characteristic association network includes: Performing redundant data filtering on the process parameters to eliminate invalid adjustment noise; Divide the core processing parameter groups based on parameter clustering algorithm and calculate the correlation between each group; Generate node interaction strength based on the association closeness and the preset influence threshold; The node interaction strength is combined with the parameter influence weight to form a multi-level influence topology graph.

[0011] Preferably, the multi-factor evaluation model includes a data compression module and a weight mapping module, and the data compression module includes: Normalizing the material deformation threshold in the key process feature to obtain a first compression vector; Performing segmented quantization encoding on the node interaction strength in the impact topology graph to generate a second compression vector; Performing a sliding mean difference calculation on the device state evolution trend, extracting state change rate characteristics, and obtaining a third compression vector; The first compressed vector, the second compressed vector and the third compressed vector are merged into a low-dimensional analysis sequence through a data fusion layer.

[0012] Preferably, the weight mapping module includes: Align the spatial dimensions of the low-dimensional analysis sequence to generate a weighted correlation matrix; Extract parameter weight features through the hierarchical attention mechanism and generate a weight distribution matrix; Perform tensor superposition operation on the weight association matrix and the weight distribution matrix to generate weight fusion features; The weighted fusion features are superimposed on the original low-dimensional analysis sequence through residual connections to output the cost prediction demand vector.

[0013] Preferably, the constructing of the adaptive prediction strategy graph by a dynamic programming algorithm includes: Initialize node attributes according to the prediction analysis module and generate edge weight matrix based on complexity weight; The cost prediction demand vector is used as the node state, and the edge weight matrix is composed of the time cost and complexity weight of the analysis sequence; Iterate the cost function to update the path value of each node and adjust the edge weight matrix; The optimal prediction analysis sequence covering all nodes is generated based on the adjusted edge weight matrix.

[0014] Preferably, the method for constructing the benchmark process parameter library includes: Collect standard process samples of various typical material deformation states to extract benchmark gradients and energy distribution thresholds; Perform multi-scale transformation on the reference gradient to generate a multi-resolution gradient model; Classify the gradient model according to the deformation category and associate it with the reference threshold database; The classified gradient model is stored as a benchmark process parameter library, and the model is regularly updated based on newly added samples.

[0015] Preferably, the parameter optimization method of the parameter clustering algorithm includes: Calculate the initial clustering granularity and minimum parameter density based on the historical processing data distribution; The parameter combinations are traversed by the grid search algorithm, and the parameters with the highest matching degree between the clustering results and the manually labeled parameter group are selected; The clustering granularity and minimum parameter density are dynamically adjusted according to the matching degree to optimize the accuracy of node interaction intensity division.

[0016] Preferably, the method for constructing the cost function includes: The cost value between nodes is defined as the balance coefficient between time cost and complexity weight; Initialize the path value of each node to zero and the starting cost value to the preset initial value; Calculate the minimum cumulative cost of each node based on the previous node through the recursive equation and record the optimal path; Generate a complete prediction analysis sequence based on forward deduction of the optimal path.

[0017] Preferably, the present invention further includes a mold stamping manufacturing cost intelligent prediction system, the system comprising: Multi-dimensional data acquisition module: used to acquire a multi-dimensional manufacturing data set, wherein the multi-dimensional manufacturing data includes material characteristic parameters, process parameters and equipment operation parameters; wherein the material characteristic parameters include material ductility coefficient and hardness distribution gradient, and the process parameters include stamping speed adjustment value and die wear index; Process feature extraction module: configured to extract key process features through process parameter analysis based on the material characteristic parameters, wherein the key process features include material deformation threshold, stress concentration factor and energy consumption fluctuation range; A topology construction module is used to generate a process influence topology map based on the process parameters through a characteristic association network construction algorithm. The influence topology map includes parameter interaction paths and node influence weights. Equipment data processing module: performs time series segmentation processing on the equipment operating parameters to generate equipment status evolution trends; Multi-factor evaluation module: inputs the key process characteristics, impact topology and equipment status evolution trend into a multi-factor evaluation model to generate a cost prediction demand vector; the multi-factor evaluation model includes a data compression module and a weight mapping module; Forecasting strategy generation module: used to construct an adaptive forecasting strategy graph based on the forecast demand vector through a dynamic programming algorithm and output a manufacturing cost forecasting plan; the nodes of the adaptive forecasting strategy graph represent the forecasting analysis modules, and the edges represent the analysis order and complexity weight.

[0018] Compared with the prior art, the present invention has the following beneficial effects: In terms of data processing, by acquiring a multi-dimensional manufacturing data set covering material characteristic parameters, process parameters, and equipment operating parameters, we comprehensively collect data on various factors that affect the manufacturing cost of mold stamping. Among them, the material ductility coefficient and hardness distribution gradient in the material characteristic parameters can accurately reflect the impact of the material's own properties on the stamping process; the stamping speed adjustment value and mold wear index in the process parameters reflect the changes in key factors in the processing process; and the equipment operating parameters record the real-time status of the equipment during the production process. This rich data lays a solid foundation for subsequent accurate analysis and cost prediction. When faced with differences in material characteristics between different batches, based on this detailed data, we can more accurately assess the impact of materials on costs and avoid cost estimation errors caused by material differences.

[0019] In the process feature extraction stage, outliers in the material characteristic parameters are eliminated to generate standardized material data, thus ensuring the reliability of the data. Based on the preset benchmark process parameter library, the gradient matching algorithm is used to identify the basic deformation categories and divide the deformation stages. The frequency domain decomposition algorithm is then used to extract the energy distribution characteristics to generate the deformation deviation, and finally the multiple key features are encoded as key process features. This processing method can deeply explore the intrinsic characteristics of the material in the stamping process and accurately grasp the change law of the material during the process. In actual production, by accurately obtaining key process characteristics such as material deformation threshold, stress concentration factor and energy consumption fluctuation range, potential problems in the process can be predicted in advance. For example, measures can be taken in advance to optimize the mold design in the stress concentration area to reduce the risk of mold damage, thereby reducing the cost of repairing and replacing the mold. At the same time, it can also improve product quality, reduce scrap rate, and indirectly reduce production costs.

[0020] When constructing the process impact topology map, redundant data on process parameters is filtered to eliminate ineffective adjustment noise. A parameter clustering algorithm is then used to group core process parameters and calculate their correlation, generating a multi-level impact topology map. This topology map clearly displays parameter interaction paths and node influence weights, providing an intuitive understanding of the complex relationships between process parameters. When adjusting the stamping speed, the topology map allows for a quick understanding of its impact on the die wear index and other related parameters, enabling the optimization of process parameter combinations, improving production efficiency, and reducing processing costs caused by inappropriate parameters.

[0021] When processing equipment operating parameters, we segment them into time series to generate equipment status evolution trends, enabling real-time tracking of changes in equipment operating status. By monitoring equipment status changes, we can proactively identify potential equipment failures and schedule maintenance in a timely manner, avoiding production interruptions caused by sudden equipment failures and reducing the economic losses associated with downtime. This also extends equipment life and reduces replacement costs.

[0022] The data compression module in the multi-factor evaluation model processes key process characteristics, relevant data in the impact topology diagram, and equipment status evolution trends, merging them into a low-dimensional analysis sequence, effectively reducing data dimensions and improving data processing efficiency while retaining key information. The weight mapping module generates a cost forecast demand vector through a series of operations, which can accurately reflect the impact of each factor on cost. This multi-factor comprehensive evaluation method greatly improves the accuracy of cost forecasts compared to traditional single-factor or simple evaluation methods. In practical applications, companies can formulate more reasonable production plans based on accurate cost forecast results, avoiding resource waste or order losses caused by cost estimation errors.

[0023] Based on the forecasted demand vector, an adaptive forecasting strategy graph is constructed using a dynamic programming algorithm to output a manufacturing cost forecast plan. The nodes and edges of this adaptive forecasting strategy graph clearly represent the forecasting analysis modules, analysis sequence, and complexity weights. Through continuous iterative optimization, an optimal forecasting analysis sequence covering all nodes is generated. This makes the forecasting process more scientific and rational, enabling dynamic adjustment of forecasting strategies based on actual production conditions to adapt to different production scenarios and data changes, further improving the reliability and practicality of cost forecasting. Based on these precise forecasting plans, companies can rationally allocate production resources, optimize process flows, and control production costs, thereby gaining a more advantageous position in the market and improving their economic efficiency and market competitiveness. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a working principle diagram of the intelligent prediction method for mold stamping manufacturing cost according to the present invention; Figure 2To generate a process impact topology flow chart; Figure 3 It is a flow chart of the data compression module; Figure 4 Flowchart of the weight mapping module. DETAILED DESCRIPTION

[0025] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0026] See also Figures 1-4 The present invention provides an intelligent prediction method for mold stamping manufacturing costs, which specifically includes the following steps: The multi-dimensional manufacturing data captured encompasses material property parameters, process parameters, and equipment operating parameters. Material property parameters include the material's ductility coefficient and hardness distribution gradient. These two parameters reflect the physical properties of the material and have a significant impact on the subsequent stamping process. Process parameters include the stamping speed adjustment value and die wear index, which are directly related to the actual operation of the stamping process. This multi-dimensional manufacturing data is collected through various sensors, data acquisition equipment, and production record systems, providing basic data support for subsequent analysis and prediction.

[0027] Based on the acquired material characteristic parameters, key process characteristics are extracted through process parameter analysis. These key process characteristics include the material deformation threshold, stress concentration factor, and energy consumption fluctuation range. The material deformation threshold determines the critical value at which plastic deformation begins during the stamping process. The stress concentration factor reflects the degree of stress concentration within the material. The energy consumption fluctuation range reflects the range of energy consumption during the stamping process. Through in-depth analysis of material characteristic parameters and the use of specific algorithms and models, these key process characteristics are extracted, providing important process basis for subsequent cost forecasting.

[0028] Based on the process parameters, a feature-association network construction algorithm is used to generate a process impact topology map. This impact topology map includes parameter interaction paths and node influence weights. The parameter interaction paths illustrate the interactions between different process parameters, while the node influence weights quantify the impact of each parameter on other parameters and the entire stamping process. Using the feature-association network construction algorithm, the process parameters are processed and analyzed to generate a process impact topology map that intuitively reflects the relationships between the parameters.

[0029] The equipment operating parameters are segmented into time series to generate equipment status evolution trends. During operation, the equipment's status changes over time. By segmenting the equipment operating parameters into time series and analyzing the characteristics and changing patterns of each data segment, the equipment status evolution trends are generated. This helps understand the changes in the equipment's operating status during the stamping process and provides equipment-related information for cost forecasting.

[0030] The extracted key process features, the generated process impact topology, and the equipment status evolution trends are input into a multi-factor evaluation model to generate a cost forecast requirement vector. This model comprehensively considers multiple factors, including process and equipment. By processing and analyzing the input data, it generates a vector that reflects the cost forecast requirements, providing critical data support for subsequent cost forecasts.

[0031] Based on the generated forecast demand vector, an adaptive forecast strategy graph is constructed using a dynamic programming algorithm to output a manufacturing cost forecast solution. The nodes of the adaptive forecast strategy graph represent forecast analysis modules, while the edges represent the analysis sequence and complexity weights. Based on the forecast demand vector, the dynamic programming algorithm combines the analysis sequence and complexity weights to construct the optimal forecast strategy graph, thereby outputting an accurate manufacturing cost forecast solution.

[0032] The present invention will be further described below in conjunction with Examples 1 to 5: Example 1: This embodiment describes in detail the specific process of extracting key process characteristics through process parameter analysis.

[0033] First, the acquired material characteristic parameters are processed to remove outliers. In the actual production process, due to sensor errors, data transmission errors, and other reasons, material characteristic parameters may have outliers. These outliers will affect the accuracy of subsequent analysis results, so they need to be removed. An outlier detection algorithm based on statistical methods, such as the 3σ criterion, is used. Assume that the material characteristic parameter set is ,in Indicates the material characteristic parameter values, calculate the mean of the set , standard deviation If a parameter value satisfy , it is determined as an outlier and removed. After the outlier removal process, standardized material data is generated.

[0034] Based on the preset reference process parameter library, the gradient matching algorithm is used to identify the basic deformation category and divide the deformation stage. The reference process parameter library stores the standard process parameters of various typical materials under different deformation states. The gradient matching algorithm identifies the basic deformation category by calculating the similarity between the gradient of the standardized material data and the reference gradient in the reference process parameter library. Assume that the standardized material data is , the base gradient is , the similarity calculation formula is ,in is the weight coefficient, which is set according to the importance of different parameters. The best-matching baseline gradient is determined based on the value of , thereby identifying the basic deformation category. Then, based on the deformation law of the material during the stamping process, the deformation stages are divided into elastic deformation stage, plastic deformation stage, etc.

[0035] The frequency domain decomposition algorithm is used to extract the energy distribution characteristics of the deformation stage, and the deformation deviation is generated by combining the reference threshold in the benchmark process parameter library. The frequency domain decomposition algorithm converts the time domain signal to the frequency domain for analysis, which can more clearly show the energy distribution of the signal. Assume that the signal in the deformation stage is , through Fourier transform Convert it to the frequency domain to get the energy distribution The reference energy distribution threshold corresponding to the deformation stage is stored in the benchmark process parameter library. , the calculation formula for deformation deviation is .

[0036] Finally, the deformation deviation, stress concentration factor, and energy consumption fluctuation range are encoded as key process characteristics. Through a specific encoding method, these characteristics are converted into a form that can be processed by computers, providing key process characteristic data for subsequent cost prediction.

[0037] Example 2: In this example, the algorithm generates a process impact topology map using a feature association network. This process plays a key role in intelligently predicting die stamping manufacturing costs. It clearly demonstrates the inherent connections between process parameters, providing an important basis for subsequent cost prediction.

[0038] The process parameters are first filtered for redundant data. In actual production, collected process parameters often contain redundant information and invalid adjustment noise. For example, minor malfunctions in equipment sensors may produce small, meaningless fluctuations in the stamping speed adjustment value, representing invalid adjustment noise. Furthermore, multiple parameters may duplicate aspects of production conditions, representing redundant data. To ensure precise analysis, a filtering method based on the similarity of data trends is employed. Process parameters such as the stamping speed adjustment value and die wear index are plotted as line graphs over a period of time, and the trends of each parameter curve are observed. If two curves show nearly identical trends for most of the time, it can be preliminarily determined that one of the parameters is redundant. The detailed changes in the parameters under different operating conditions are then compared to identify and eliminate redundant parameters. This processing ensures that the data for subsequent analysis is more representative and accurate.

[0039] After completing the redundant data filtering, the parameter clustering algorithm is used to divide the core processing parameter groups. Here, the hierarchical clustering algorithm is selected, which can gradually merge and form clusters of different levels according to the similarity between parameters, just like building blocks. Taking the parameters such as the stamping speed adjustment value, mold temperature, and mold wear index as an example, each parameter is first regarded as a separate class. The similarity between different parameters is calculated, for example, using the Euclidean distance to measure, the formula is , in Represents two parameter vectors and The smaller the distance, the more similar the two parameters. Each time, the two clusters with the closest distance are merged to form a new cluster. This process is repeated until the preset number of clusters is reached or the distance between clusters no longer meets the merging conditions. In this way, highly correlated parameters can be grouped together to form a core processing parameter group.

[0040] Next, the degree of correlation between each group is calculated. For the divided core processing parameter groups, the degree of correlation is determined by analyzing their changes in different production batches. For example, one group contains the stamping speed adjustment value and the die wear index, and the other group contains the die temperature and stamping pressure. Across multiple production batches, if it is found that when the stamping speed adjustment value in the first group increases, the die temperature in the second group also shows a clear upward trend, then this indicates that there is a high degree of correlation between the two sets of parameters. Mutual information is used to quantify this degree of correlation; the greater the mutual information, the closer the correlation between the two sets of parameters.

[0041] The node interaction strength is generated based on the correlation strength and a preset influence threshold. This threshold is determined based on extensive historical production data and actual production experience. If the correlation strength exceeds this threshold, it indicates that the two sets of parameters have a significant mutual influence, and a corresponding node interaction strength needs to be generated. For example, if the influence threshold is set to 0.5, when the calculated mutual information between the two sets of parameters is 0.6, the condition is met. In this case, the node interaction strength is determined based on the difference between the correlation strength and the threshold. The larger the difference, the higher the interaction strength.

[0042] Finally, the node interaction strength is combined with the parameter influence weight to form a multi-level influence topology diagram. The parameter influence weight is determined through expert evaluation and actual production data feedback. For example, based on experience, experts determine that the stamping speed adjustment value has a greater impact on the mold stamping manufacturing cost and assign it a higher weight. The node interaction strength and parameter influence weight are combined and presented in a graphical form. Nodes are used to represent each core processing parameter group, and edges are used to represent the interaction relationship between them. The thickness of the edge indicates the size of the node interaction strength, and the color of the edge indicates the high or low parameter influence weight. This forms a multi-level influence topology diagram that intuitively displays the complex relationship between process parameters.

[0043] Example 3: This embodiment focuses on the data compression module in the multi-factor evaluation model, with the aim of converting complex data such as key process characteristics, impact topology diagrams, and equipment status evolution trends into a low-dimensional analysis sequence that is easier to handle, thereby improving the efficiency and accuracy of subsequent cost forecasting.

[0044] First, the material deformation threshold in the key process feature is normalized to obtain the first compression vector. The material deformation threshold varies greatly under different materials and stamping processes. Taking two different mold materials as an example, the deformation threshold of one material may be in the range of 5-10, and the other may be in the range of 50-100. In order to unify the data scale, the normalization method is used. Assume that the material deformation threshold is , its value range is , the normalized formula is ,here This is the normalized result. The normalized material deformation thresholds are arranged in a certain order to form the first compression vector. For example, after normalization, the values obtained are 0.3, 0.5, 0.7, etc., which are arranged in this order to form a vector.

[0045] The node interaction strengths in the impact topology graph are segmented and quantized to generate a second compressed vector. The node interaction strengths in the impact topology graph are continuously changing values, so segmented quantization is required for easier computer processing. Based on experience and data distribution characteristics, the range of node interaction strengths is divided into several intervals. For example, the five intervals are [0, 0.2), [0.2, 0.4), [0.4, 0.6), [0.6, 0.8), and [0.8, 1]. For each node interaction strength value, determine which interval it belongs to and assign the corresponding code. If the node interaction strength value is 0.3, it is encoded as 1; if it is 0.7, it is encoded as 3. All the encoded values of the node interaction strengths are arranged in order to obtain the second compressed vector.

[0046] Perform sliding mean difference calculation on the device state evolution trend, extract the state change rate characteristics, and obtain the third compression vector. The device state changes over time, and its change rate characteristics can be effectively extracted by sliding mean difference calculation. Set a sliding window, and the window size is selected based on the characteristics and experience of the device operation data. Assume that 5 time points are taken. For the values of the device operation parameters in the time series , first calculate the sliding mean. Take these 5 values as an example, the sliding mean As time passes, the window slides backward by one time point, and a new sliding mean is calculated. The sliding mean difference, the difference between two adjacent sliding means, is then calculated. These sliding mean differences are arranged in chronological order to form a third compressed vector, which reflects the rate of change of the device state.

[0047] Finally, the data fusion layer combines the first, second, and third compressed vectors into a low-dimensional analysis sequence. The data fusion layer uses a simple concatenation method to connect the three compressed vectors in sequence. For example, if the first compressed vector is [0.3, 0.5, 0.7], the second compressed vector is [1, 2, 3], and the third compressed vector is [0.1, -0.2, 0.3], the resulting low-dimensional analysis sequence is [0.3, 0.5, 0.7, 1, 2, 3, 0.1, -0.2, 0.3], providing processing data for the subsequent weight mapping module.

[0048] Example 4: This embodiment details the weight mapping module in the multi-factor evaluation model, performs spatial dimension alignment on the low-dimensional analysis sequence, and generates a weight correlation matrix. The low-dimensional analysis sequence is formed by merging multiple compressed vectors. Different compressed vectors may have different dimensions and characteristics. In order to facilitate weight analysis, spatial dimension alignment is required. Assume that the low-dimensional analysis sequence is , by filling zeros or performing dimension transformation, etc., so that each element has the same dimension. Then, based on the correlation between different elements, a weighted correlation matrix is generated ,For example , in Indicates the The first element eigenvalues, Indicates the The mean of the elements, is the number of features.

[0049] The parameter weight features are extracted through the hierarchical attention mechanism to generate a weight distribution matrix. The hierarchical attention mechanism can assign different weights according to the importance of different elements in the low-dimensional analysis sequence. Assume that the low-dimensional analysis sequence is , the first layer of attention mechanism calculates the attention weight of each element ,in is calculated through a neural network and the element The second-layer attention mechanism further processes the attention weights of the first layer to obtain the final weight distribution matrix .

[0050] Perform tensor superposition operation on the weight association matrix and the weight distribution matrix to generate weight fusion features. The tensor superposition operation can fuse the information of the two matrices to obtain more comprehensive weight information. Assume that the weight association matrix is , the weight distribution matrix is , weight fusion feature ,in Represents a tensor superposition operation.

[0051] The weight fusion feature is superimposed on the original low-dimensional analysis sequence through residual connection to output the cost prediction demand vector. Residual connection can avoid the loss of information in the weight mapping process and ensure the accuracy of the model. Assume that the weight fusion feature is , the original low-dimensional analysis sequence is , cost forecast demand vector ,In this way, the cost forecast demand vector is output, providing key data support for subsequent cost forecast.

[0052] Example 5: This embodiment mainly describes the specific process of constructing an adaptive prediction strategy graph through a dynamic programming algorithm, initializing node attributes according to the prediction analysis module, and generating an edge weight matrix based on the complexity weight. The prediction analysis module is a node in the adaptive prediction strategy graph, and each node has different attributes, such as the function of the node, computational complexity, etc. When initializing the node attributes, they are set according to the specific functions and characteristics of the prediction analysis module. The complexity weight reflects the computational complexity of each prediction analysis module and is set according to actual conditions. The edge weight matrix represents the connection weight between nodes, and the edge weight is composed of the time cost of the analysis sequence and the complexity weight. Assume that the node and nodes The edge weight between , the time cost is , the complexity weight is ,but .

[0053] The cost forecast demand vector is used as the node state, and the edge weight matrix is composed of the time cost and complexity weights of the analysis sequence. The cost forecast demand vector contains information on various aspects, such as process and equipment. Using it as the node state provides input data for each forecast analysis module. The edge weight matrix determines the connection relationships and weights between nodes, affecting the order and complexity of the forecast analysis.

[0054] The path value of each node is updated iteratively through the cost function, and the edge weight matrix is adjusted. The cost function is used to measure the path cost from the starting point to each node, and the cost value between nodes is defined as the balance coefficient of time cost and complexity weight. Assume that the balance coefficient is ,node To Node The cost value Initialize the path value of each node to zero and the starting cost value to the preset initial value (for example ). Through the recursive equation Calculate the minimum cumulative cost of each node based on the previous node and record the optimal path, where Representation node The minimum cumulative cost of Representation node According to the calculated minimum cumulative cost, the edge weight matrix is adjusted to make the edge weight more in line with the actual prediction requirements.

[0055] Based on the adjusted edge weight matrix, an optimal forecast analysis sequence covering all nodes is generated. This optimal forecast analysis sequence covering all nodes is obtained by finding the optimal path from the starting point to the end point. Graph search algorithms such as the Dijkstra algorithm can be used to find the optimal path. Based on this optimal path, the execution order of the forecast analysis modules is determined, resulting in an accurate manufacturing cost forecast solution.

[0056] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0057] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A method for intelligent prediction of die stamping manufacturing cost, characterized in that: include: Acquire a multi-dimensional manufacturing data set; the multi-dimensional manufacturing data includes material characteristic parameters, process parameters, and equipment operating parameters; the material characteristic parameters include material ductility coefficient and hardness distribution gradient, and the process parameters include stamping speed adjustment value and die wear index; Based on the material characteristic parameters, extracting key process characteristics through process parameter analysis, wherein the key process characteristics include material deformation threshold, stress concentration factor and energy consumption fluctuation range; According to the process parameters, a process influence topology map is generated by a characteristic association network construction algorithm, wherein the influence topology map includes parameter interaction paths and node influence weights; Processing the equipment operating parameters in time series segments to generate equipment status evolution trends; Inputting the key process characteristics, impact topology and equipment status evolution trend into a multi-factor evaluation model to generate a cost prediction demand vector; Based on the predicted demand vector, an adaptive prediction strategy graph is constructed through a dynamic programming algorithm to output a manufacturing cost prediction solution; the nodes of the adaptive prediction strategy graph represent prediction analysis modules, and the edges represent analysis order and complexity weight.

2. The intelligent prediction method for mold stamping manufacturing cost according to claim 1, characterized in that: The key process characteristics are extracted through process parameter analysis, including: Eliminating abnormal values from the material characteristic parameters to generate standardized material data; Based on the preset benchmark process parameter library, the basic deformation categories are identified through the gradient matching algorithm and the deformation stages are divided; The frequency domain decomposition algorithm is used to extract the energy distribution characteristics of the deformation stage, and the deformation deviation is generated by combining the reference threshold in the benchmark process parameter library; The deformation deviation, stress concentration factor and energy consumption fluctuation range are encoded as key process characteristics.

3. The intelligent prediction method for mold stamping manufacturing cost according to claim 1, characterized in that: The process impact topology map is generated by constructing an algorithm using a characteristic association network, including: Performing redundant data filtering on the process parameters to eliminate invalid adjustment noise; Divide the core processing parameter groups based on parameter clustering algorithm and calculate the correlation between each group; Generate node interaction strength based on the association closeness and the preset influence threshold; The node interaction strength is combined with the parameter influence weight to form a multi-level influence topology graph.

4. The intelligent prediction method for mold stamping manufacturing cost according to claim 1, characterized in that: The multi-factor evaluation model includes a data compression module and a weight mapping module, and the data compression module includes: Normalizing the material deformation threshold in the key process feature to obtain a first compression vector; Performing segmented quantization encoding on the node interaction strength in the impact topology graph to generate a second compression vector; Performing a sliding mean difference calculation on the device state evolution trend, extracting state change rate characteristics, and obtaining a third compression vector; The first compressed vector, the second compressed vector and the third compressed vector are merged into a low-dimensional analysis sequence through a data fusion layer.

5. The intelligent prediction method for mold stamping manufacturing cost according to claim 4, characterized in that: The weight mapping module includes: Align the spatial dimensions of the low-dimensional analysis sequence to generate a weighted correlation matrix; Extract parameter weight features through the hierarchical attention mechanism and generate a weight distribution matrix; Perform tensor superposition operation on the weight association matrix and the weight distribution matrix to generate weight fusion features; The weighted fusion features are superimposed on the original low-dimensional analysis sequence through residual connections to output the cost prediction demand vector.

6. The intelligent prediction method for mold stamping manufacturing cost according to claim 1, characterized in that: The method of constructing an adaptive prediction strategy graph by a dynamic programming algorithm includes: Initialize node attributes according to the prediction analysis module and generate edge weight matrix based on complexity weight; The cost prediction demand vector is used as the node state, and the edge weight matrix is composed of the time cost and complexity weight of the analysis sequence; Iterate the cost function to update the path value of each node and adjust the edge weight matrix; The optimal prediction analysis sequence covering all nodes is generated based on the adjusted edge weight matrix.

7. The intelligent prediction method for mold stamping manufacturing cost according to claim 2, characterized in that: The method for constructing the benchmark process parameter library includes: Collect standard process samples of various typical material deformation states to extract benchmark gradients and energy distribution thresholds; Perform multi-scale transformation on the reference gradient to generate a multi-resolution gradient model; Classify the gradient model according to the deformation category and associate it with the reference threshold database; The classified gradient model is stored as a benchmark process parameter library, and the model is regularly updated based on newly added samples.

8. The intelligent prediction method for mold stamping manufacturing cost according to claim 3, characterized in that: The parameter optimization method of the parameter clustering algorithm includes: Calculate the initial clustering granularity and minimum parameter density based on the historical processing data distribution; The parameter combinations are traversed by the grid search algorithm, and the parameters with the highest matching degree between the clustering results and the manually labeled parameter group are selected; The clustering granularity and minimum parameter density are dynamically adjusted according to the matching degree to optimize the accuracy of node interaction intensity division.

9. The intelligent prediction method for mold stamping manufacturing cost according to claim 6, characterized in that: The method for constructing the cost function includes: The cost value between nodes is defined as the balance coefficient between time cost and complexity weight; Initialize the path value of each node to zero and the starting cost value to the preset initial value; Calculate the minimum cumulative cost of each node based on the previous node through the recursive equation and record the optimal path; Generate a complete prediction analysis sequence based on forward deduction of the optimal path.

10. An intelligent prediction system for mold stamping manufacturing costs, characterized in that: include: Multi-dimensional data acquisition module: used to acquire a multi-dimensional manufacturing data set, wherein the multi-dimensional manufacturing data includes material characteristic parameters, process parameters and equipment operation parameters; wherein the material characteristic parameters include material ductility coefficient and hardness distribution gradient, and the process parameters include stamping speed adjustment value and die wear index; Process feature extraction module: configured to extract key process features through process parameter analysis based on the material characteristic parameters, wherein the key process features include material deformation threshold, stress concentration factor and energy consumption fluctuation range; A topology construction module is used to generate a process influence topology map based on the process parameters through a characteristic association network construction algorithm. The influence topology map includes parameter interaction paths and node influence weights. Equipment data processing module: performs time series segmentation processing on the equipment operating parameters to generate equipment status evolution trends; Multi-factor evaluation module: inputs the key process characteristics, impact topology and equipment status evolution trend into a multi-factor evaluation model to generate a cost prediction demand vector; the multi-factor evaluation model includes a data compression module and a weight mapping module; Forecasting strategy generation module: used to construct an adaptive forecasting strategy graph based on the forecast demand vector through a dynamic programming algorithm and output a manufacturing cost forecasting plan; the nodes of the adaptive forecasting strategy graph represent the forecasting analysis modules, and the edges represent the analysis order and complexity weight.

Images