Precision mold structure parameterization rapid design system based on multi-distance data fusion

Abstract

The invention relates to the technical field of computer aided design, and discloses a precise mold structure parameterization rapid design system based on multi-source data fusion. The system comprises a data acquisition module, a time sequence alignment module, a parameter calculation module, a structure generation module and a verification feedback module, time sequence alignment and fusion are carried out on multi-source heterogeneous data, a multi-target genetic algorithm is adopted to optimize mold key parameters in parallel, and parameter closed-loop correction is realized based on virtual combined simulation. According to the method, the consistency, precision and reliability of mold structure design can be effectively improved.

Description

Technical Field

[0001] This invention belongs to the field of computer-aided design technology, specifically relating to a rapid parametric design system for precision mold structures based on multi-source data fusion. Background Technology

[0002] In the fields of precision manufacturing and digital design, the application of computer-aided design technology and data-driven methods has become a core means to improve product development efficiency and quality. Among them, precision molds, as key process equipment for achieving high-precision part forming, directly determine the final product's forming quality and production cost through the accuracy of their structural design and their rapid response capability.

[0003] Parametric rapid design of precision mold structures is a key technological focus in the industry, aiming to achieve efficient generation and optimization of mold structures through parametric modeling and intelligent algorithms. The fundamental goal of this technology is to transform design experience and process constraints into calculable parametric models, thereby shortening the design cycle and improving the reliability of design results.

[0004] Existing technologies typically rely on multi-source data to determine and verify mold structural parameters. However, there are significant time-asynchrony issues in the fusion process of multi-source data, which leads to deviations in the calculation of structural parameters, thereby causing micro-deformation of the mold cavity and resulting in a decrease in mold closing accuracy.

[0005] Existing systems lack a unified time-series alignment mechanism when processing heterogeneous data, making it difficult to achieve effective coordination between design parameters and real-time process data, further exacerbating the uncertainty of parameter optimization. The contradiction between the accuracy requirements of mold structure design and the insufficient reliability of multi-source data fusion has become a key challenge restricting the development of rapid design technology for precision molds. Summary of the Invention

[0006] The technical problem to be solved by the present invention is to overcome the defects of existing precision mold parametric design systems, which are caused by the asynchronous timing of multi-source data, resulting in deviations in the calculation of structural parameters, and thus causing micro-deformation of mold cavity and a decrease in mold closing accuracy. The present invention provides a rapid parametric design system for precision mold structure based on multi-source data fusion.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: The precision mold structure parameterized rapid design system based on multi-source data fusion includes a data acquisition module, a timing alignment module, a parameter calculation module, a structure generation module, and a verification feedback module.

[0008] The data acquisition module is used to collect heterogeneous data in real time from the mold design history database, online detection sensor group, process execution monitoring unit and material property library.

[0009] The timing alignment module is connected to the data acquisition module and is used to apply a unified timestamp to the acquired multi-source heterogeneous data and perform dynamic alignment using a sliding window to eliminate timing asynchronous errors between data.

[0010] The parameter calculation module is connected to the timing alignment module and is used to perform parallel optimization calculations of mold structure parameters based on the aligned fused data.

[0011] The structure generation module is connected to the parameter calculation module, which is used to automatically construct a three-dimensional parametric model of the mold based on the optimized parameters.

[0012] The verification feedback module is connected to the structure generation module and the data acquisition module respectively. It is used to perform virtual simulation of the generated 3D model and compare the simulation results with the measured data to correct the parameter calculation strategy.

[0013] Furthermore, the timing alignment module includes a timestamp unit and a window alignment unit. The timestamp unit assigns an absolute time stamp based on the GPS clock to each input data stream. The window alignment unit uses a sliding time window of 2 seconds to interpolate and resample each data stream within the window, ensuring that all data are strictly synchronized on the time axis.

[0014] Furthermore, the parameter calculation module includes a parameter initialization unit, a constraint loading unit, and a parallel optimization unit. The parameter initialization unit retrieves an initial parameter set from a standard template library based on the mold type. The constraint loading unit extracts geometric, mechanical, and thermal constraints from the process execution monitoring unit and the material property library. The parallel optimization unit employs a multi-objective genetic algorithm to simultaneously optimize the cavity dimensions, gating system layout, and cooling channel arrangement parameters while satisfying all constraints.

[0015] Furthermore, the objective functions of the multi-objective genetic algorithm include minimizing cavity volume error, maximizing filling balance, and maximizing cooling uniformity. The algorithm population size is set to 128, the crossover probability is 0.85, the mutation probability is 0.02, and the iteration termination condition is that the improvement of the optimal solution is less than 1 / 1000 for 10 consecutive generations.

[0016] Furthermore, the structure generation module includes a feature recognition unit and a model building unit. The feature recognition unit automatically recognizes the mold base structure, cavity features, and auxiliary mechanisms based on the optimized parameters output by the parameter calculation module. The model building unit calls the parametric modeling kernel based on the recognition results and generates a complete three-dimensional solid model through feature stitching and Boolean operations.

[0017] Furthermore, the verification feedback module includes a virtual mold closing unit and a deviation analysis unit. The virtual mold closing unit simulates the closing process of the moving and fixed molds in a simulation environment, detecting interference and gap distribution. The deviation analysis unit calculates the root mean square error between the simulated gap and the actual sensor-detected gap. When the error exceeds 5 micrometers, it triggers the constraint weight adaptive adjustment mechanism of the parameter solution module.

[0018] Furthermore, the constraint weight adaptive adjustment mechanism dynamically adjusts the weight coefficients of the corresponding constraints in the multi-objective genetic algorithm according to the type of deviation. For mold clearance deviations caused by micro-deformation of the cavity, the weight of geometric constraints is increased; for deformations caused by thermal stress, the weight of thermal constraints is increased. The weight adjustment step size is 0.1, and the maximum adjustment range is twice the initial weight.

[0019] Furthermore, the system adopts a hierarchical control architecture, including a decision-making layer, an execution layer, and a sensing layer. The decision-making layer is responsible for parameter calculation and verification strategy formulation; the execution layer is responsible for 3D model generation and simulation analysis; and the sensing layer is responsible for multi-source data acquisition and preprocessing. The three layers interact with each other through a standard data interface to ensure the real-time performance and consistency of the system.

[0020] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention effectively eliminates the asynchronous timing error in the multi-source data fusion process through the timing alignment module, providing a highly consistent data foundation for parameter calculation; 2. The parameter calculation module adopts a multi-objective parallel optimization algorithm to simultaneously determine the key structural parameters of the mold under strict constraints, which significantly reduces the risk of micro-deformation of the cavity caused by insufficient parameter coupling; 3. The verification feedback module compares the virtual mold fitting simulation with the measured data to achieve closed-loop correction of the design parameters, further improving the mold fitting accuracy and mold structure reliability; 4. The system adopts a hierarchical architecture, which ensures efficient collaboration and stable operation throughout the entire process from data acquisition to model generation. Attached Figure Description

[0021] Figure 1 This is a schematic diagram of the overall technical solution architecture of the precision mold structure parameterized rapid design system based on multi-source data fusion proposed in this invention; Figure 2 This is a schematic diagram of the core principle framework of the timing alignment module and parameter calculation module in this invention; Figure 3 This is a flowchart of the multi-objective parallel optimization logic of the parameter calculation module in this invention. Figure 4 This is a logical flow diagram of the structure generation module and the verification feedback module in this invention; Figure 5 This is a schematic diagram of the multi-level interaction relationship and data flow between the decision-making layer, execution layer and sensing layer in this invention. Detailed Implementation

[0022] Example 1: Please refer to the appendix Figure 1 This embodiment details the technical implementation of a rapid parameterized design system for precision mold structures based on multi-source data fusion. The system aims to solve the problem of calculation deviations in mold structure parameters caused by asynchronous timing of multi-source data through a highly integrated modular architecture, ultimately achieving suppression of micro-deformation in the mold cavity and improvement of mold closing accuracy. The core components of the system include a data acquisition module, a timing alignment module, a parameter calculation module, a structure generation module, and a verification feedback module. These modules are seamlessly connected and operate collaboratively through standard data interfaces.

[0023] The data acquisition module serves as the system's data input terminal, responsible for real-time acquisition of heterogeneous data from four main data sources.

[0024] The primary data source is a historical database of mold designs. This database stores the mold structure parameters, material formulas, and process settings of past successful cases. The data is organized in the form of a structured query language table and accessed through a database connection interface.

[0025] The second data source is an online detection sensor group, which is integrated and deployed on mold processing machine tools and mold testing equipment. It includes a laser displacement sensor, an infrared thermal imager, and a vibration accelerometer, which are used to monitor the mold cavity size, surface temperature distribution, and mechanical vibration status in real time.

[0026] The laser displacement sensor has a measurement accuracy of ±0.1 micrometers and a sampling frequency of 1 kilohertz; the infrared thermal imager has a temperature resolution of 0.1 degrees Celsius and a spatial resolution of 320 x 240 pixels; the vibration acceleration meter has a range of -50g to +50g and a frequency response range of 0 Hz to 5 kilohertz.

[0027] The third data source is the process execution monitoring unit, which is embedded in the control system of the injection molding machine or die casting machine. It continuously collects process parameters such as injection pressure, holding time, melt temperature and screw speed. The data is transmitted through the fieldbus at a rate of 100 data points per second.

[0028] The fourth data source is the material properties library, which contains mechanical property data, coefficients of thermal expansion, and creep curves of various mold steels and engineering plastics. The data is stored in a standardized material card format and supports indexing and querying by material grade and heat treatment status.

[0029] The data acquisition module integrates a data buffer and a format converter. It performs preliminary validity checks on all input data, removes outliers that are significantly beyond the physical range, and encapsulates heterogeneous data into data packets with source and time identifiers. These packets are then temporarily stored in a first-in-first-out queue, awaiting invocation by the timing alignment module.

[0030] The timing alignment module is directly connected to the output of the data acquisition module. Its core function is to eliminate timing asynchrony errors between multi-source data, providing time consistency assurance for subsequent parameter calculation. Please refer to the attached document. Figure 2 The timing alignment module consists of two processing stages: a timestamp marking unit and a window alignment unit.

[0031] After receiving data packets from the data acquisition module, the timestamp unit immediately appends an absolute time stamp based on the GPS clock to each data packet. This unit incorporates a high-precision clock synchronization circuit, whose clock source is locked to the GPS satellite signal, ensuring the global uniqueness and microsecond-level accuracy of the timestamp.

[0032] The window alignment unit employs a sliding time window mechanism with a fixed length of 2 seconds to dynamically align the data streams within the window. Specifically, this unit first divides the 2-second window into 2000 equally spaced time slices, each corresponding to 1 millisecond, based on the system master clock.

[0033] For each data stream, the window alignment unit detects the distribution of data points within the window. If a data stream is missing data in a specific time slice, a cubic spline interpolation algorithm is used to resample based on neighboring data points to generate interpolated data for that time slice. If a data stream has multiple data points in a specific time slice, the data point with the timestamp closest to the center of that time slice is taken as the representative value. After interpolation and resampling, all data streams have a synchronized data point in each 1-millisecond time slice within the 2-second window, thus achieving strict alignment of multi-source data on the time axis.

[0034] The aligned data is reassembled into fused data frames in time slice order. Each data frame contains fields such as time slice number, sensor readings, process parameters, and material properties, and is transmitted to the parameter calculation module via a high-speed data bus.

[0035] The parameter calculation module receives the fused data frame output by the timing alignment module and performs parallel optimization calculations of the mold structure parameters. Please refer to the appendix. Figure 2 With appendix Figure 3The parameter calculation module is internally divided into three functional sub-modules: a parameter initialization unit, a constraint loading unit, and a parallel optimization unit. The parameter initialization unit retrieves the corresponding initial parameter set from a pre-set standard template library based on the type identifier of the mold to be designed. The standard template library is categorized by mold type, such as two-plate molds, three-plate molds, and hot runner molds. Each template contains initial values ​​for parameters such as basic cavity dimensions, preliminary gating system layout, and default cooling channel routing.

[0036] The constraint loading unit simultaneously extracts geometric, mechanical, and thermal constraints from the fused data frame. Geometric constraints are derived from the laser displacement data of the online detection sensor group and tolerance standards in the design history database, specifically including the minimum cavity wall thickness, draft angle range, and parting surface flatness requirements.

[0037] Mechanical constraints are derived from yield strength data in the material property library and injection pressure history from the process execution monitoring unit, mainly involving the stress safety factor and fatigue life cycle of the mold components under maximum injection pressure. Thermal constraints are based on temperature field data from infrared thermal imagers and the thermal expansion coefficient of materials, limiting the maximum thermal deformation of the mold and the lower limit of the cooling rate within the operating temperature range.

[0038] The parallel optimization unit employs a multi-objective genetic algorithm to simultaneously optimize three key structural parameters—cavity size, gating system layout, and cooling channel arrangement—while satisfying all the aforementioned constraints. The objective function of this algorithm is defined as a weighted combination of minimizing cavity volume error, maximizing filling balance, and maximizing cooling uniformity.

[0039] The cavity volume error refers to the absolute difference between the designed cavity volume and the theoretical product volume; the filling balance is quantified by the reciprocal of the time difference between each gate reaching the end of the cavity; and the cooling uniformity is characterized by the negative standard deviation of the temperature distribution on the cavity surface. The algorithm population size is set to 128 individuals, and each individual is encoded as a set of structural parameter vectors, including 35 optimization variables such as cavity length, width, and height, gate position coordinates, cooling channel diameter and spacing.

[0040] In the genetic operations, the crossover probability is set to 0.85, using a simulated binary crossover operator; the mutation probability is set to 0.02, using a polynomial mutation operator. During the iteration process, each generation of the population undergoes selection, crossover, and mutation to produce offspring, and the objective function value and constraint violation degree of each individual are calculated. The algorithm terminates when the improvement of the objective function of the optimal solution for 10 consecutive generations is less than 0.1%, i.e., the relative change rate of the objective function value of the optimal solution between adjacent generations is less than 0.001. After optimization, the parallel optimization unit outputs a combination of structural parameters that makes the three objective functions optimal and satisfies all constraints. This result is transmitted as the optimization parameter set to the structure generation module.

[0041] The structure generation module automatically constructs a 3D parametric model of the mold based on the optimized parameter set passed from the parameter calculation module. Please refer to the attached diagram. Figure 4 The structure generation module consists of a feature recognition unit and a model building unit connected in series. The feature recognition unit first parses and optimizes the parameter set, classifying and mapping it to the three major feature categories of the mold: mold base structure features, cavity features, and auxiliary mechanism features.

[0042] The mold base structure features include template dimensions, guide pillar and guide sleeve positions, and reset rod layout; cavity features cover the product forming surface geometry, slider core-pulling mechanism, and venting groove distribution; auxiliary mechanism features involve hot runner system, ejection device, and cooling water connector. The feature recognition unit, based on a predefined feature template library, instantiates a corresponding parameterized feature object for each recognized feature and establishes parent-child dependencies and geometric constraints between features.

[0043] The model building unit, based on feature recognition results, calls the embedded parametric modeling kernel to gradually generate a complete 3D solid model through feature stitching and Boolean operations. The modeling kernel uses boundary representation for geometric description and supports feature-based modeling operations such as extrusion, rotation, sweep, and lofting. The specific modeling process is as follows: First, based on the structural feature parameters of the mold base, three-dimensional solids of the main body plates, such as the moving template and the fixed template, are generated; then, the cavity features are cut out from the main body plates through negative Boolean operations to form the product cavity; Next, based on the gating system layout parameters, create channel structures such as runners and gates on the template; Based on the cooling water channel layout parameters, a cooling water circuit is constructed inside the template, ensuring that the water channel maintains a constant distance from the cavity surface; Finally, auxiliary mechanism features are integrated, such as the assembly of ejector plates and the installation of hot runner nozzles. All feature operations maintain parametric correlation, meaning that when any value in the optimization parameter set changes, the model building unit can automatically trigger the update and reconstruction of related features, ensuring geometric consistency of the model. The generated 3D solid model is output in a standard graphic format, along with a complete feature tree and parameter history for subsequent verification and modification.

[0044] The verification feedback module receives the 3D model generated by the structure generation module and performs a virtual simulation comparison with the measured data to achieve closed-loop correction of the design parameters. Please refer to the appendix. Figure 4 The verification feedback module includes a virtual mold closing unit and a deviation analysis unit. The virtual mold closing unit simulates the closing process of the moving and fixed molds under the action of closing forces within an integrated finite element simulation environment.

[0045] After the simulation model is imported into the 3D solid model, it is automatically meshed into tetrahedral meshes, with the number of meshes typically between 1 million and 5 million to ensure calculation accuracy. Simulation boundary conditions include the magnitude of the clamping force, the motion curve of the clamping mechanism, the elastic modulus of the template, and the friction coefficient of the contact surface. These parameters are extracted from the fused data of the process execution monitoring unit and the material property library.

[0046] The virtual mold-closing unit performs nonlinear static analysis to calculate the contact pressure distribution between the moving and fixed molds, the elastic deformation of the mold template, and the microscopic gap changes in the cavity region during the mold-closing process. The simulation results are visualized as contour plots showing the interference region and gap size, and the maximum gap value, average gap value, and gap distribution uniformity index are output.

[0047] The deviation analysis unit synchronously receives measured data of the mold closing gap collected by the online detection sensor group during the actual mold trial. This unit calculates the root mean square error (RMSE) between the gap distribution obtained from the virtual mold closing simulation and the actual sensor-detected gap. The specific formula for calculating the RMS error is as follows: ; in The total number of testing sites. For the first Simulated gap values ​​at each detection point This represents the measured gap value at the corresponding point. The deviation analysis unit has a preset error threshold of 5 micrometers. When the calculated root mean square error exceeds this threshold, the constraint weight adaptive adjustment mechanism within the parameter solution module is immediately triggered.

[0048] This adjustment mechanism dynamically corrects the weight coefficients of corresponding constraints in the multi-objective genetic algorithm based on the type of deviation. The specific strategy is as follows: If the deviation analysis indicates that the mold closing clearance deviation is mainly caused by micro-deformation of the cavity, the weight coefficient of the geometric constraint in the objective function is increased; if the deviation is mainly caused by template deformation due to thermal stress, the weight coefficient of the thermal constraint is increased. The weight adjustment is incremented or decremented in steps of 0.1, but the maximum adjustment range is limited to no more than twice the initial weight to prevent oscillations in the optimization process. The adjusted weight coefficients are immediately fed back to the parallel optimization unit of the parameter solution module, driving it to re-execute the parameter optimization calculation, thus forming a closed-loop iteration of design-simulation-comparison-correction until the root mean square error is reduced to below 5 micrometers, ensuring that the final generated mold structure parameters meet the high-precision mold closing requirements.

[0049] Please refer to the attached document. Figure 5This system adopts a hierarchical control architecture, encompassing three levels: the decision-making layer, the execution layer, and the sensing layer. The sensing layer, at the bottom, consists of a data acquisition module and its connected databases and sensors, responsible for acquiring and pre-processing raw data. The execution layer, in the middle, includes the timing alignment module, the structure generation module, and the virtual modeling unit within the verification feedback module, responsible for specific tasks such as data synchronization and alignment, 3D model construction, and simulation analysis. The decision-making layer, at the top, comprises the parameter calculation module and the deviation analysis unit within the verification feedback module, responsible for intelligent decision-making functions such as parameter optimization strategy formulation and constraint weight adjustment.

[0050] The three layers exchange information through standardized data interfaces. The interface protocols define strict data formats, transmission rates, and handshake signals to ensure that instructions and data are transmitted efficiently and without errors between layers, guaranteeing the real-time performance and global consistency of the system.

[0051] Example 2: This example provides another implementation of a rapid parametric design system for precision mold structures based on multi-source data fusion, focusing on alternative technical solutions for its timing alignment module and parameter calculation module in specific application scenarios. Please refer to the appendix. Figure 1 With appendix Figure 2 The system architecture of this embodiment is consistent with that of Embodiment 1, but there are significant differences in the data processing flow and optimization algorithm configuration to adapt to design environments with extremely high real-time requirements or limited computing resources.

[0052] In the timing alignment module, this embodiment employs a dynamic alignment strategy based on event triggering rather than a fixed time window. The timestamp marking unit still assigns a GPS clock mark to the input data, but the window alignment unit no longer uses a fixed 2-second sliding window. Instead, this unit monitors the data arrival events of each data stream in real time. When any critical data stream, such as a laser displacement sensor or injection pressure sensor, generates a new data point, an alignment event is immediately triggered. During the alignment event, the window alignment unit uses the timestamp of the triggered data point as a reference and collects the latest data points from all other data streams within a very short time tolerance band, for example, ±10 milliseconds.

[0053] For data streams that fail to arrive in time within the tolerance band, the system no longer performs interpolation. Instead, it directly uses the last valid data point from the data stream to fill the gap and marks that point as the predicted value. This event-triggered mechanism significantly reduces the computational latency and memory overhead of the alignment process, making it particularly suitable for industrial environments with large differences in data stream rates or unstable network transmission latency. The aligned data is also organized into fused data frames, but the frame structure becomes event-driven, with each frame header containing the trigger event type, timestamp, and data integrity flag.

[0054] In this embodiment, the parameter calculation module has replaced the algorithm for the parallel optimization unit. Given the inherent limitations of multi-objective genetic algorithms in terms of iterative convergence speed, this embodiment uses a decomposition-based multi-objective evolutionary algorithm as the core optimizer. This algorithm decomposes the original multi-objective optimization problem into a series of scalarized sub-problems, and each sub-problem aggregates multiple objective functions into a single objective through a weighted summation method. Specifically, for the three objectives of cavity volume error, filling balance, and cooling uniformity, the algorithm defines a set of uniformly distributed weight vectors, with each weight vector corresponding to a sub-problem.

[0055] Individuals in the population are assigned to different subproblems for optimization, and comparisons between individuals are only performed within the same subproblem. The algorithm uses a differential evolution strategy for individual updates, with the crossover probability adjusted to 0.9 and the mutation probability adjusted to 0.01 to enhance global exploration capabilities. The population size is set to 100, less than the 128 in Example 1, to balance computational efficiency and solution set diversity. The iteration termination condition is changed to a maximum of 200 iterations to avoid failing to output results within a limited time due to overly strict convergence criteria.

[0056] This decomposition-based optimization strategy typically exhibits faster convergence speeds when dealing with high-dimensional parameter spaces, making it particularly suitable for online design applications requiring rapid response. After optimization, the algorithm outputs a set of Pareto optimal solutions rather than a single solution, allowing design engineers to select the most suitable combination of structural parameters from the solution set based on their preferences, thus providing greater flexibility in the design process.

[0057] In this embodiment, the structure generation module enhances the feature recognition capability of the model building unit. In addition to the basic feature recognition described in Embodiment 1, the feature recognition unit in this embodiment also integrates a rule-based intelligent feature recommendation mechanism. This mechanism automatically learns the commonly used feature combinations and layout patterns of different types of molds by analyzing successful cases in the historical design database.

[0058] Upon receiving the optimized parameter set, the feature recognition unit not only instantiates the features directly corresponding to the parameters, but also recommends additional enhanced features based on the rule base. For example, it automatically suggests adding reinforcing ribs in the high-pressure injection area and recommends adding vent plugs in the deep cavity structure.

[0059] These recommended features are provided to the model building unit as options, and designers can confirm or reject their inclusion in the final model. This retains the interface for human experience intervention on the basis of automation, thereby improving the practicality and reliability of the design solution.

[0060] In this embodiment, the verification feedback module enhances the multi-source data fusion verification capability of the deviation analysis unit. The simulation analysis performed by the virtual mold closing unit is no longer limited to the mold closing gap, but extends to the simulation of the entire mold operation process, including stages such as melt filling, pressure holding and cooling, and mold opening and ejection. Accordingly, the deviation analysis unit needs to compare more types of simulation data with measured data, such as the melt front position at the end of filling, the temperature field changes during the cooling process, and the product displacement during the ejection stage.

[0061] The root mean square error calculation is therefore extended to a multi-index comprehensive error, obtained by weighted averaging of the errors of key physical quantities at each stage. The error thresholds are also set differently according to the importance of the indexes. For example, the mold closing gap threshold is kept at 5 micrometers, the filling balance error threshold is set at 3%, and the cooling uniformity error threshold is set at 2 degrees Celsius.

[0062] When any index error exceeds the limit, the deviation analysis unit will trigger constraint weight adjustment. However, the adjustment strategy is more refined. It will dynamically calculate and adjust the weights of the corresponding objective function and constraints in multi-objective optimization based on the specific type of the index exceeding the limit and its sensitivity relationship with each structural parameter, so as to achieve more precise closed-loop feedback control.

[0063] In this embodiment, the overall hierarchical architecture of the system further optimizes the data compression and caching strategies between layers. Given the larger data volume and higher real-time requirements in this embodiment, the sensing layer adds a wavelet transform-based data compression algorithm during data preprocessing to perform lossy compression on the original sensor data, reducing the amount of data transmitted to the upper layers.

[0064] The interface between the execution layer and the decision-making layer introduces a data caching mechanism. For non-urgent batch data, such as query results from the historical case library, it can be temporarily stored in the cache and retrieved on demand, avoiding blocking the real-time data stream. These optimizations ensure that the system can maintain stable performance and response speed even under resource-constrained or high-load conditions.

Claims

1. A rapid parametric design system for precision mold structures based on multi-source data fusion, characterized in that, include: The data acquisition module is used to collect heterogeneous data in real time from the mold design history database, online detection sensor group, process execution monitoring unit and material property library; The timing alignment module, connected to the data acquisition module, is used to apply a unified timestamp to the acquired multi-source heterogeneous data and perform dynamic alignment using a sliding window to eliminate timing asynchronous errors between data. The parameter calculation module, connected to the timing alignment module, is used to perform parallel optimization calculations of mold structure parameters based on the aligned fused data. The structure generation module is connected to the parameter calculation module and is used to automatically construct a three-dimensional parametric model of the mold based on the optimized parameters. The verification feedback module is connected to the structure generation module and the data acquisition module respectively. It is used to perform virtual simulation of the generated 3D model and compare the simulation results with the measured data to correct the parameter calculation strategy.

2. The precision mold structure parameterized rapid design system based on multi-source data fusion according to claim 1, characterized in that, The timing alignment module includes a timestamp marking unit and a window alignment unit; The timestamp unit assigns an absolute time stamp based on the GPS clock to each input data stream; The window alignment unit uses a sliding time window of 2 seconds to interpolate and resample each data stream within the window, ensuring that all data are strictly synchronized on the time axis.

3. The precision mold structure parameterized rapid design system based on multi-source data fusion according to claim 1, characterized in that, The parameter calculation module includes a parameter initialization unit, a constraint loading unit, and a parallel optimization unit. The parameter initialization unit calls the initial parameter set from the standard template library according to the mold type; The constraint loading unit extracts geometric constraints, mechanical constraints, and thermal constraints from the process execution monitoring unit and the material property library; The parallel optimization unit uses a multi-objective genetic algorithm to simultaneously optimize the cavity size, gating system layout, and cooling water channel arrangement parameters while satisfying all constraints.

4. The precision mold structure parameterized rapid design system based on multi-source data fusion according to claim 3, characterized in that, The objective functions of the multi-objective genetic algorithm include minimizing cavity volume error, maximizing filling balance, and maximizing cooling uniformity. The algorithm population size was set to 128, the crossover probability was 0.85, the mutation probability was 0.02, and the iteration termination condition was that the improvement of the optimal solution was less than 1 / 1000 for 10 consecutive generations.

5. The precision mold structure parameterized rapid design system based on multi-source data fusion according to claim 1, characterized in that, The structure generation module includes a feature recognition unit and a model building unit; The feature recognition unit automatically identifies the mold base structure, cavity features, and auxiliary mechanisms based on the optimized parameters output by the parameter calculation module; The model building unit calls the parametric modeling kernel based on the recognition results, and generates a complete 3D solid model through feature stitching and Boolean operations.

6. The precision mold structure parameterized rapid design system based on multi-source data fusion according to claim 1, characterized in that, The verification feedback module includes a virtual model-combining unit and a deviation analysis unit; The virtual mold closing unit simulates the closing process of the moving and fixed molds in a simulation environment, and detects interference and gap distribution; The deviation analysis unit calculates the root mean square error between the simulated gap and the actual sensor detection gap. When the error exceeds 5 micrometers, the constraint weight adaptive adjustment mechanism of the parameter solution module is triggered.

7. The precision mold structure parameterized rapid design system based on multi-source data fusion according to claim 6, characterized in that, The constraint weight adaptive adjustment mechanism dynamically adjusts the weight coefficients of the corresponding constraints in the multi-objective genetic algorithm according to the type of deviation. For mold clearance deviations caused by micro-deformation of the cavity, increase the weight of geometric constraints; for deformations caused by thermal stress, increase the weight of thermal constraints. The weight adjustment step size is 0.1, and the maximum adjustment range is twice the initial weight.

8. The precision mold structure parameterized rapid design system based on multi-source data fusion according to claim 1, characterized in that, The system adopts a hierarchical control architecture, including a decision layer, an execution layer, and a sensing layer; the decision layer is responsible for parameter calculation and verification strategy formulation. The execution layer is responsible for 3D model generation and simulation analysis; the sensing layer is responsible for multi-source data acquisition and preprocessing. The three layers exchange information through standard data interfaces to ensure the real-time performance and consistency of the system.

9. The precision mold structure parameterized rapid design system based on multi-source data fusion according to claim 3, characterized in that, The parallel optimization unit uses a multi-objective genetic algorithm to perform parallel optimization calculations as follows: Initialize a population set of 128 individuals; perform a simulated binary crossover operation with a crossover probability of 0.85; Perform a polynomial mutation operation with a mutation probability of 0.02; Evaluate the objective function value and constraint violation degree of each individual; terminate the iteration when the improvement of the optimal solution is less than 1 / 1000 for 10 consecutive generations.

10. The precision mold structure parameterized rapid design system based on multi-source data fusion according to claim 6, characterized in that, The process by which the deviation analysis unit calculates the root mean square error is as follows: Obtain the total number of detection points N; Calculate the squared difference between the simulated gap value and the measured gap value at each detection point; Sum the squared differences of all detection points and divide by the total number of detection points; The root mean square error is obtained by taking the square root of the result. When the root mean square error exceeds 5 micrometers, the constraint weight adaptive adjustment mechanism is triggered.

Images