Magnesium alloy die casting process quality on-line regulation system

Abstract

The application discloses a magnesium alloy die casting process quality online regulation system, comprising a injection state acquisition module, a data preprocessing module, a melt front recognition module, a mold thermal memory generation module, a flow-heat coupling mismatch judgment module, a defect risk vector generation module, a parameter contribution degree inversion module and a collaborative regulation execution module. The system acquires current die injection state data and combines historical data to generate process data, historical thermal memory values, standard arrival time and standard process data; generates melt front disturbance index, actual arrival time and thermal memory matrix; calculates flow-heat coupling mismatch index with the target mold area as the index, generates a defect risk vector and inverts the priority regulation parameters, generates the safety regulation instruction of the current die and / or the next die within the safety boundary, and realizes the die casting quality online regulation.

Description

Technical Field

[0001] This invention relates to the field of metal pressure casting and intelligent control technology for die casting processes, and particularly to an online quality control system for magnesium alloy die casting processes. Background Technology

[0002] Magnesium alloys are characterized by low density, high specific strength, good vibration damping performance, and good thermal conductivity, and are widely used in automotive parts, electronic device housings, lightweight structural components, and other fields. Magnesium alloy die casting is one of the important processes for the mass production of magnesium alloy parts, and it typically includes steps such as melting and holding, molten metal filling, injection filling, pressurization and holding, cooling and solidification, mold opening and ejection, and spray cooling.

[0003] In the die casting process of magnesium alloys, the quality of the castings is affected by a variety of factors, including injection conditions, cavity pressure, vacuum conditions, mold temperature, cooling conditions, coating conditions, and production cycle. Because magnesium alloy melts have a low density and a strong tendency to oxidize, and thin-walled and complex structural areas are particularly sensitive to filling and solidification conditions, fluctuations in process parameters, abnormal local mold temperatures, or unstable venting can easily lead to quality problems in the castings, such as cold shuts, incomplete casting, gas entrapment, oxide inclusions, shrinkage cavities, hot cracks, sticking to the mold, and dimensional fluctuations.

[0004] In current die-casting production, the die-casting process is typically controlled by setting injection curves, mold temperature ranges, vacuum parameters, cooling parameters, and spraying parameters. However, under continuous production conditions, the various processes influence each other, and fluctuations in a single parameter can cause multiple quality problems. The forming state of different areas can also change with production cycle time, mold thermal equilibrium, and material conditions. Therefore, relying solely on preset process parameters or single-item condition monitoring is insufficient to accurately reflect the actual forming state of the current die.

[0005] Furthermore, some quality anomalies are delayed and cumulative, often only being discovered after casting is formed through visual inspection, dimensional inspection, airtightness testing, or non-destructive testing. By the time the inspection results are fed back to the production process, the anomaly may have already affected multiple consecutive batches, leading to increased scrap rates and reduced process adjustment efficiency. Simultaneously, different defects may be interrelated; adjusting a single process parameter may improve the risk of one defect but increase the risk of another.

[0006] Therefore, magnesium alloy die casting production still needs a technical solution that can adapt to continuous die production processes, comprehensively utilize die casting process status data, and adjust process parameters online based on quality risks, in order to improve the stability of the die casting process and the consistency of casting quality. Summary of the Invention

[0007] To address the shortcomings of existing technologies, the present invention aims to provide an online quality control system for magnesium alloy die casting processes. This system integrates injection state data and historical data to identify regional mismatch risks between the melt filling state and the mold thermal state, and determines priority control parameters accordingly to achieve online coordinated control of die casting quality.

[0008] To achieve the above objectives, the present invention provides the following technical solution: an online quality control system for magnesium alloy die casting process, comprising: The injection status acquisition module is used to acquire the injection status data of the current cycle. The data preprocessing module is connected to the injection state acquisition module. It is used to perform time alignment, segmentation and normalization on the injection state data and the pre-stored historical data to obtain the target mold area, the current mold process data, historical thermal memory values, standard arrival time and standard process data. The melt front identification module, connected to the data preprocessing module, is used to generate the melt front disturbance index and the actual arrival time of each target mold area based on the injection pressure change rate, injection speed change rate, and the deviation between the cavity pressure arrival time and the standard arrival time in the current mold process data. The mold thermal memory generation module, connected to the data preprocessing module, is used to generate a thermal memory matrix based on the mold zone temperature and historical thermal memory values. The fluid-thermal coupling mismatch judgment module connects the melt front identification module and the mold thermal memory generation module. It is used to calculate the fluid-thermal coupling mismatch index by coupling the melt front disturbance index, actual arrival time and thermal memory matrix with the target mold area as the index. The defect risk vector generation module, connected to the flow-thermal coupling mismatch judgment module, is used to generate defect risk vectors for multiple types of defect risks based on the flow-thermal coupling mismatch index. The parameter contribution inversion module connects the defect risk vector generation module and the data preprocessing module. It is used to determine the priority control parameters based on the defect risk vector, the deviation of the current process parameters from the standard process data, and the mechanism correlation coefficient between defect risk and process parameters. The collaborative control execution module, connected to the parameter contribution inversion module, is used to generate safety control instructions for the current and / or next mode within a preset safety boundary based on the priority control parameters, so as to adjust at least one of the injection, pressurization, vacuum, cooling, and spraying parameters.

[0009] Furthermore, the injection status acquisition module includes: The injection curve acquisition unit is used to acquire the injection pressure, injection speed and injection rod displacement of the current module and generate injection curve data; The cavity status acquisition unit is used to acquire the cavity pressure, vacuum pressure, and boost pressure of the current mold and generate cavity response data; The mold thermal state acquisition unit is used to acquire the temperature of the mold zone in the current mold batch and generate zone thermal state data; The mold number identification unit is connected to the injection curve acquisition unit, the cavity status acquisition unit and the mold thermal status acquisition unit respectively. It is used to associate the injection curve data, the cavity response data and the partition thermal status data according to the current mold number to generate injection status data with mold number identification. The data preprocessing module includes: The time synchronization unit, connected to the mold identification unit, is used to time-align the injection status data with the mold identification and the pre-stored historical data based on the injection rod displacement characteristic point, the cavity pressure rise point, or the pressurization pressure rise point to obtain synchronized status data. The working condition segmentation unit, connected to the time synchronization unit, is used to divide the synchronization state data into a low-speed injection segment, a fast filling segment, and a pressurized solidification segment according to the characteristics of the injection speed change and the characteristics of the cavity pressure change, so as to obtain segmented state data. The region mapping unit, connected to the working condition segmentation unit, is used to map the mold partition temperature in the segmented state data to each of the target mold regions, and associate the cavity pressure arrival signal determined by the cavity response data with the corresponding target mold region to obtain regionalized process data; The normalized baseline generation unit, connected to the region mapping unit, is used to generate the current mold process data, the historical thermal memory value, the standard arrival time, and the standard process data based on the regionalized process data and the pre-stored historical data. The current mold process data and the standard arrival time are output to the melt front identification module, and the historical thermal memory value is output to the mold thermal memory generation module.

[0010] Furthermore, the melt front identification module includes: The leading edge signal extraction unit, connected to the data preprocessing module, is used to extract from the current process data the injection pressure, injection speed, cavity pressure, and melt leading edge sensing signal formed by the injection state acquisition module and the data preprocessing module. The melt leading edge sensing signal includes at least one of cavity pressure arrival signal, ultrasonic echo signal, and electromagnetic induction signal. The arrival time determination unit is connected to the leading edge signal extraction unit and is used to determine the actual arrival time of the melt leading edge to each of the target mold areas based on the cavity pressure rise point or the melt leading edge sensing signal change point corresponding to each of the target mold areas. The disturbance feature generation unit, connected to the arrival time determination unit, is used to generate leading edge disturbance feature data based on the rate of change of the injection pressure, the rate of change of the injection velocity, the deviation between the actual arrival time and the standard arrival time, and the fluctuation amplitude of the melt leading edge sensing signal. The disturbance index generation unit, connected to the disturbance feature generation unit, is used to normalize and weighted fuse the leading edge disturbance feature data to generate the melt leading edge disturbance index; The disturbance type determination unit, connected to the disturbance index generation unit, is used to determine the type of melt front disturbance based on the comparison result of the melt front disturbance index and the preset disturbance threshold, combined with the actual arrival time difference between each target mold area, the vacuum pressure fluctuation characteristics extracted from the current mold process data, and the continuity of cavity pressure rise. The type of melt front disturbance includes at least one of front lag, front split, air entrapment disturbance, backflow disturbance, and impact filling. The unit outputs the melt front disturbance index, the type of melt front disturbance, and the actual arrival time to the flow-thermal coupling mismatch determination module.

[0011] Furthermore, the mold thermal memory generation module includes: The area temperature extraction unit is connected to the data preprocessing module and is used to extract the mold partition temperature corresponding to each target mold area from the current mold process data to obtain the current mold partition temperature data. The historical thermal memory retrieval unit is connected to the data preprocessing module and is used to retrieve the historical thermal memory value and retrieve the standard thermal memory value and thermal memory decay coefficient corresponding to each target mold area from the standard process data to obtain thermal memory reference data. The thermal memory recursion unit is connected to the area temperature extraction unit and the historical thermal memory recall unit, respectively. It is used to recursively calculate the thermal accumulation state of each target mold area based on the current module partition temperature data, the historical thermal memory value and the thermal memory decay coefficient, and generate the thermal memory matrix. The thermal memory matrix includes the thermal memory value corresponding to each target mold area. A thermal state determination unit, connected to the thermal memory recursion unit, is used to generate thermal state data for each target mold area based on the deviation between each thermal memory value and the standard thermal memory value, and in combination with a preset cooling threshold, a preset thermal stability threshold, and a preset thermal accumulation threshold. The thermal memory matrix output unit is connected to the thermal state determination unit and is used to output the thermal memory matrix and the thermal state data to the thermal coupling mismatch determination module.

[0012] Furthermore, the thermal-fluid coupling mismatch determination module includes: The region index association unit is connected to the melt front identification module and the mold thermal memory generation module respectively. It is used to associate the actual arrival time of the melt front to each target mold area, the melt front disturbance index, and the thermal memory value of the corresponding target mold area in the thermal memory matrix with each target mold area as an index to obtain region flow thermal association data. A temperature gradient generation unit, connected to the data preprocessing module and the region index association unit, is used to generate a normalized temperature gradient for each target mold region based on the mold partition temperature in the current mold process data. The mismatch feature generation unit, connected to the region index association unit and the temperature gradient generation unit, is used to generate region mismatch feature data based on the region flow-heat association data, the deviation between the actual arrival time and the standard arrival time, the deviation between the thermal memory value and the standard thermal memory value obtained from the standard process data, and the normalized temperature gradient. The mismatch index generation unit, connected to the mismatch feature generation unit, is used to perform weighted fusion of the regional mismatch feature data to generate the flow-thermal coupling mismatch index corresponding to each target mold region, and output the flow-thermal coupling mismatch index to the defect risk vector generation module.

[0013] Furthermore, the thermal-fluid coupling mismatch determination module also includes: The mismatch level determination unit, connected to the mismatch index generation unit, is used to compare the flow-thermal coupling mismatch index corresponding to each of the target mold areas with a preset flow-thermal mismatch threshold to generate the flow-thermal mismatch level of each target mold area; The mismatch cause classification unit, connected to the mismatch level determination unit, is used to generate flow-thermal mismatch cause data based on the flow-thermal mismatch level, the melt front disturbance index, the deviation between the actual arrival time and the standard arrival time, the deviation between the thermal memory value and the standard thermal memory value, the normalized temperature gradient, and the cavity pressure change characteristics extracted from the current mold process data. The flow-thermal mismatch cause data includes at least one of cold-state hysteresis mismatch, front disturbance mismatch, arrival time mismatch, thermal accumulation mismatch, and solidification pressure mismatch. The mismatch result output unit is connected to the mismatch cause classification unit and is used to output the flow-heat coupling mismatch index, the flow-heat mismatch level and the flow-heat mismatch cause data to the defect risk vector generation module, so that the defect risk vector generation module generates a corresponding defect risk vector based on the flow-heat mismatch cause data.

[0014] Furthermore, the injection state acquisition module is also used to acquire the demolding force data and spraying state feedback data of the current mold cycle. The data preprocessing module is also used to perform time alignment and normalization processing on the demolding force data and the spraying state feedback data, so that the processed demolding force data and spraying state feedback data are incorporated into the current mold cycle process data. The defect risk vector generation module includes: The mismatch information receiving unit is connected to the thermal coupling mismatch judgment module. It is used to receive the thermal coupling mismatch index corresponding to each target mold area, and extract the vacuum pressure fluctuation characteristics, solidification stage cavity pressure change characteristics, mold partition temperature change characteristics, demolding force change characteristics and spraying status feedback characteristics from the current mold process data to obtain the basic data for defect judgment. The regional defect mapping unit, connected to the mismatch information receiving unit, is used to associate the target mold area with at least one of the following risk types according to the location attributes of each target mold area and the flow-heat coupling mismatch index, thereby obtaining regional defect mapping data. The risk component generation unit, connected to the region defect mapping unit, is used to generate, respectively, cold shut risk component, air entrapment risk component, oxidation inclusion risk component, shrinkage cavity or shrinkage porosity risk component, hot crack risk component and mold sticking risk component based on the region defect mapping data, the flow-heat coupling mismatch index and the defect judgment basic data. The risk level determination unit is connected to the risk component generation unit and is used to compare each risk component with the corresponding preset defect risk threshold to generate the defect risk level corresponding to each risk component. The risk vector output unit, connected to the risk level determination unit, is used to combine the cold shut risk component, the air entrapment risk component, the oxidation inclusion risk component, the shrinkage cavity or shrinkage porosity risk component, the hot crack risk component, and the mold sticking risk component into the defect risk vector, and output the defect risk vector and the defect risk level to the parameter contribution inversion module.

[0015] Furthermore, the parameter contribution inversion module includes: A risk component receiving unit, connected to the defect risk vector generation module, is used to receive the defect risk vector and extract various defect risk components therein. The parameter deviation generation unit is connected to the data preprocessing module and the collaborative control execution module, respectively. It is used to extract the current process parameters from the current module process data and the actuator status data of the previous or current module executed by the collaborative control execution module, and generate a process parameter deviation vector according to the difference between the current process parameters and the standard process data. The current process parameters include at least one of the following: low-speed injection speed, fast-slow speed switching point, second speed, boost pressure, boost delay time, vacuum valve opening time, zone cooling intensity, and zone spraying amount. The mechanism association calling unit is connected to the parameter deviation generation unit and is used to call the mechanism association coefficient between defect risk and process parameters from the standard process data or pre-stored mechanism association data to form mechanism association data; An execution effectiveness retrieval unit, connected to the data preprocessing module, is used to retrieve the adjustment effectiveness of each actuator on the corresponding process parameters from the standard process data or pre-stored historical data, forming execution effectiveness data; The contribution calculation unit is connected to the risk component receiving unit, the parameter deviation generation unit, the mechanism association calling unit, and the execution effectiveness calling unit, respectively, and is used to calculate the parameter contribution corresponding to each current process parameter based on the defect risk vector, the process parameter deviation vector, the mechanism association data, and the execution effectiveness data.

[0016] Furthermore, the parameter contribution inversion module also includes: A contribution ranking unit, connected to the contribution calculation unit, is used to rank the contribution of each parameter to obtain a parameter contribution ranking result. A dominant parameter determination unit, connected to the contribution ranking unit, is used to determine at least one dominant process parameter based on the parameter contribution ranking result, and to determine the dominant process parameter as the priority control parameter. When the difference between the highest parameter contribution and the second highest parameter contribution is less than a preset contribution difference threshold, the corresponding multiple process parameters are determined as the priority control parameters. The control direction generation unit is connected to the dominant parameter determination unit. It is used to generate control direction data for reducing the corresponding defect risk based on the defect risk component corresponding to the priority control parameter, the process parameter deviation vector and the mechanism correlation coefficient, and generate a strong correction flag when the parameter contribution is greater than a preset strong correction threshold. The inversion result output unit is connected to the control direction generation unit and is used to output the priority control parameters, the control direction data and the strong correction identifier to the collaborative control execution module.

[0017] Furthermore, the coordinated control execution module includes: The control strategy matching unit is connected to the parameter contribution inversion module and is used to receive the priority control parameters and the control direction data, and determine the adjustable parameters of the current module and the correction parameters of the next module based on the defect risk component corresponding to the priority control parameters and the control direction data. The current module rapid closed-loop unit is connected to the control strategy matching unit. When the current module is still within the preset adjustable time window, it generates a rapid closed-loop control command based on the adjustable parameters of the current module. The rapid closed-loop control command is used to adjust at least one of the following: low-speed injection speed, fast-slow speed switching point, second speed, pressurization delay time, and vacuum valve opening time. The next-mode slow closed-loop unit is connected to the control strategy matching unit and is used to generate a slow closed-loop control command based on the next-mode correction parameters. The slow closed-loop control command is used to correct at least one of the following in the next-mode: boost pressure, zone cooling intensity, zone spraying amount, and initial injection curve. A safety boundary verification unit is connected to the current module fast closed-loop unit and the next module slow closed-loop unit, respectively. It is used to compare the fast closed-loop control command and the slow closed-loop control command with the preset safety boundary, and limit or replace the corresponding control command with a suboptimal control command when the corresponding safety boundary threshold is exceeded, so as to obtain the safety control command. The control command output unit is connected to the safety boundary verification unit and is used to output the safety control command to at least one of the following: injection controller, booster controller, vacuum valve controller, cooling controller or cooling water proportional valve, and spraying actuator.

[0018] The beneficial effects of this invention are: This invention collects current injection state data and combines it with pre-stored historical data for time alignment, segmentation, and normalization, thereby generating unified and traceable current process data and standard process data. By generating a melt front disturbance index based on the injection pressure change rate, injection speed change rate, and cavity pressure arrival time deviation, it can more accurately reflect the actual filling state of magnesium alloy melt. By generating a thermal memory matrix based on the mold zone temperature and historical thermal memory values, it is possible to characterize the thermal accumulation state of different areas of the mold during continuous production. By using the target mold area as an index, the melt front disturbance index, actual arrival time, and thermal memory matrix are coupled and calculated to obtain the flow-thermal coupling mismatch index, thereby improving the accuracy of defect formation precursor identification. By generating a defect risk vector based on the flow-heat coupling mismatch index and determining the priority control parameters by combining process parameter deviations and mechanism correlation coefficients, the targeting of process regulation can be improved. By generating safety control commands for the current and / or next die within preset safety boundaries, coordinated control of injection, pressurization, vacuum, cooling, and spraying parameters can be achieved, thereby improving the stability of the magnesium alloy die casting process and the consistency of casting quality. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the online quality control system for the magnesium alloy die-casting process in this invention; Figure 2 This is a schematic diagram of the structure of the injection state acquisition module in this invention; Figure 3 This is a schematic diagram of the data preprocessing module in this invention; Figure 4 This is a schematic diagram of the melt front identification module in this invention; Figure 5 This is a schematic diagram of the mold thermal memory generation module in this invention; Figure 6 This is a schematic diagram of the thermal coupling mismatch judgment module in this invention; Figure 7 This is a schematic diagram of the defect risk vector generation module in this invention; Figure 8 This is a schematic diagram of the parameter contribution inversion module in this invention; Figure 9 This is a schematic diagram of the structure of the collaborative control execution module in this invention.

[0020] Figure reference numerals: 1. Injection state acquisition module; 11. Injection curve acquisition unit; 12. Cavity state acquisition unit; 13. Mold thermal state acquisition unit; 14. Mold batch identification unit; 2. Data preprocessing module; 21. Time synchronization unit; 22. Working condition segmentation unit; 23. Region mapping unit; 24. Normalized benchmark generation unit; 3. Melt front identification module; 31. Front signal extraction unit; 32. Arrival time determination unit; 33. Disturbance feature generation unit; 34. Disturbance index generation unit; 35. Disturbance type determination unit; 4. Mold thermal memory generation module; 41. Region temperature extraction unit; 42. Historical thermal memory recall unit; 43. Thermal memory recursion unit; 44. Thermal state determination unit; 45. Thermal memory matrix output unit; 5. Fluid-thermal coupling mismatch judgment module; 51. Region index association unit; 52. Temperature gradient generation unit; 53. Mismatch feature generation unit; 54. 55. Mismatch Index Generation Unit; 56. Mismatch Level Determination Unit; 57. Mismatch Cause Classification Unit; 58. Mismatch Result Output Unit; 6. Defect Risk Vector Generation Module; 69. Mismatch Information Receiving Unit; 60. Regional Defect Mapping Unit; 61. Risk Component Generation Unit; 62. Risk Level Determination Unit; 63. Risk Vector Output Unit; 74. Parameter Contribution Inversion Module; 75. Risk Component Receiving Unit; 76. Parameter Deviation Generation Unit; 77. Mechanism Association Calling Unit; 78. Execution Effectiveness Calling Unit; 79. Contribution Calculation Unit; 70. Contribution Ranking Unit; 71. Dominant Parameter Determination Unit; 71. Control Direction Generation Unit; 82. Inversion Result Output Unit; 83. Coordinated Control Execution Module; 84. Control Strategy Matching Unit; 85. Current Module Fast Closed-Loop Unit; 86. Next Module Slow Closed-Loop Unit; 87. Safety Boundary Verification Unit; 88. Control Command Output Unit. Detailed Implementation

[0021] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0022] Example 1, as Figure 1 As shown, this embodiment provides an online quality control system for magnesium alloy die casting, suitable for continuous die casting production of thin-walled magnesium alloy shells, supports, cover plates, connectors, and lightweight structural parts. The die casting material can be AZ91D magnesium alloy or other magnesium alloy materials suitable for die casting. The die casting equipment can be a cold chamber die casting machine. The mold can be divided into a gate area, runner area, thin-walled area, thick-walled area, rib area, overflow area, and ejection area according to forming function and defect-sensitive areas. Each area serves as a target mold area for subsequent data processing.

[0023] The system includes an injection state acquisition module 1, a data preprocessing module 2, a melt front identification module 3, a mold thermal memory generation module 4, a flow-thermal coupling mismatch judgment module 5, a defect risk vector generation module 6, a parameter contribution inversion module 7, a collaborative control execution module 8, and a data feedback update module.

[0024] The injection status acquisition module 1 is used to collect injection status data for the current die casting cycle. This data includes injection pressure, injection speed, injection rod displacement, cavity pressure, vacuum pressure, booster pressure, mold zone temperature, demolding force data, and spraying status feedback data. Through this data acquisition, the system can obtain the flow status, pressure status, thermal status, demolding status, and spraying execution status during the die casting process. For the specific hardware configuration, sampling frequency, and data processing method of the injection status acquisition module 1, please refer to Example 2.

[0025] Data preprocessing module 2 is connected to injection state acquisition module 1 and is used to perform time alignment, working condition segmentation, region mapping, and normalization processing on injection state data and pre-stored historical data to obtain target mold area, current mold process data, historical thermal memory values, standard arrival time, and standard process data. The specific implementation of normalization processing can adopt a nonlinear saturated normalization function, or an equivalent normalization method that can convert data of different dimensions into dimensionless deviation values.

[0026] The melt front identification module 3 is connected to the data preprocessing module 2. It generates a melt front disturbance index and the actual arrival time of the melt front to each target mold area based on the injection pressure change rate, injection speed change rate, deviation between the cavity pressure arrival time and the standard arrival time, and the fluctuation amplitude of the melt front sensing signal in the current mold process data. The melt front sensing signal includes at least one of the cavity pressure arrival signal, ultrasonic echo signal, and electromagnetic induction signal. When no ultrasonic echo sensor or electromagnetic induction sensor is configured, the system uses the cavity pressure arrival signal as the basic melt front sensing signal. Therefore, the system does not determine the melt position solely based on the injection rod stroke, but rather comprehensively considers the pressure response, speed change, and area arrival time to determine the actual melt filling state.

[0027] The mold thermal memory generation module 4 is connected to the data preprocessing module 2 and is used to generate a thermal memory matrix based on the mold zone temperature of the current mold batch and historical thermal memory values. The thermal memory matrix is ​​used to characterize the thermal accumulation, thermal stability, or cooling state of each target mold area during continuous multi-mold production. Compared with single mold temperature detection, the thermal memory matrix can reflect the cumulative trend of thermal states across multiple mold batches, thereby identifying areas where the current temperature has not yet significantly exceeded the limit but where thermal drift or localized thermal accumulation already exists.

[0028] The flow-thermal coupling mismatch judgment module 5 is connected to the melt front identification module 3 and the mold thermal memory generation module 4, respectively. It uses the target mold region as an index to couple the melt front disturbance index, actual arrival time, thermal memory value, and temperature gradient to generate a flow-thermal coupling mismatch index corresponding to each target mold region. This module is used to determine whether the thermal state of a target mold region is suitable for magnesium alloy melt filling and solidification when the melt front reaches that region. This judgment method avoids misjudgments caused by judging melt flow anomalies or mold temperature anomalies in isolation.

[0029] The defect risk vector generation module 6 connects to the fluid-thermal coupling mismatch judgment module 5. It generates a defect risk vector based on the fluid-thermal coupling mismatch index, fluid-thermal mismatch cause data, and vacuum pressure fluctuation characteristics, solidification stage cavity pressure change characteristics, mold zone temperature change characteristics, demolding force change characteristics, and spraying status feedback characteristics in the current mold process data. The defect risk vector includes cold shut risk, air entrapment risk, oxidation inclusion risk, shrinkage cavity or porosity risk, thermal cracking risk, and mold sticking risk. Through vectorization, multiple defect risks can be output simultaneously, avoiding the inability to guide subsequent adjustments by only outputting a single alarm result.

[0030] The parameter contribution inversion module 7 connects to the defect risk vector generation module 6 and the data preprocessing module 2. It is used to calculate the parameter contribution of each process parameter to the current defect risk based on the defect risk vector, the deviation of the current process parameters from the standard process data, the mechanistic correlation coefficient between defect risk and process parameters, and the effectiveness of actuator adjustment, and to determine the priority control parameters. The current process parameters include at least one of the following: low-speed injection speed, fast / slow speed switching point, second-speed speed, boost pressure, boost delay time, vacuum valve opening time, zone cooling intensity, and zone coating volume.

[0031] The coordinated control execution module 8 connects to the parameter contribution inversion module 7 and is used to generate safety control commands for the current and / or next module within a preset safety boundary based on priority control parameters and control direction data. The current module's fast closed-loop is mainly used to adjust the low-speed injection speed, fast / slow speed switching point, second-speed speed, pressurization delay time, and vacuum valve opening time; the next module's slow closed-loop is mainly used to adjust the pressurization pressure, zone cooling intensity, zone spraying volume, and initial injection curve. The safety control commands can be output to the injection controller, pressurization controller, vacuum valve controller, cooling controller, cooling water proportional valve, or spraying actuator.

[0032] The data feedback and update module is used to feed back die-casting process data, actuator feedback data, demolding force data, demolding thermal imaging data, and subsequent inspection results to the system, updating standard process data, defect risk assessment rules, mechanism correlation coefficients, execution effectiveness, and initial process parameters for the next mold. Subsequent inspection results can include appearance inspection results, dimensional inspection results, X-ray inspection results, airtightness inspection results, weight inspection results, and mechanical property sampling inspection results.

[0033] The working principle of this embodiment is as follows: The system first collects the current die-casting state data and combines it with pre-stored historical data to generate current die-casting process data, historical thermal memory values, standard arrival time, and standard process data. Then, it identifies the melt front disturbance state and actual arrival time, and generates a thermal memory matrix for the target die region. Next, using the target die region as an index, it determines the degree of mismatch between the melt flow state and the die thermal state. Following this, it generates a defect risk vector based on the mismatch results and inversely identifies process parameters that significantly contribute to defect risk. Finally, within the safety boundary, it outputs safety control instructions for the current or next die, and updates subsequent judgments and control criteria through data feedback. Through this process, the system can identify precursors to defect formation under continuous production conditions and transform defect risks into executable process control actions, thereby improving the stability of the magnesium alloy die-casting process and the consistency of casting quality.

[0034] Example 2, as Figure 2-6 As shown, this embodiment, based on Embodiment 1, further explains the specific structure, data processing method, and algorithm implementation of the injection state acquisition module 1, data preprocessing module 2, melt front identification module 3, mold thermal memory generation module 4, and flow-thermal coupling mismatch judgment module 5. This embodiment is used to realize injection state acquisition, data preprocessing, melt front identification, mold thermal memory generation, and flow-thermal coupling mismatch judgment, solving the problems of different data sampling frequencies, unclear mold-time correspondence, unclear target mold area correspondence, difficulty in directly identifying melt front state, difficulty in quantifying continuous multi-mold thermal state, and difficulty in coupling and judging flow state and thermal state in multi-source die casting processes.

[0035] In this embodiment, z represents only the target mold region number, Z represents the total number of target mold regions; r represents only the disturbance feature number; u represents only the flow-heat mismatch cause category number; ℓ and h are only used as summation subscripts. The output values ​​of each normalization function are dimensionless, and the inputs of each exponential function and sigmoid function are dimensionless. The sigmoid function is defined as: ; in, This is the output value of the Sigmoid function. Input is a dimensionless function.

[0036] 1. Injection status acquisition module 1; The injection status acquisition module 1 is used to acquire real-time status data during the magnesium alloy die casting process. The injection status acquisition module 1 includes an injection curve acquisition unit 11, a cavity status acquisition unit 12, a mold thermal status acquisition unit 13, and a mold number identification unit 14.

[0037] The injection curve acquisition unit 11 is used to acquire the injection pressure, injection speed, and injection rod displacement of the current cycle and generate injection curve data. The injection rod displacement sensor can be installed at the injection cylinder or the injection rod movement mechanism to acquire the injection rod displacement; the injection speed can be directly acquired by the injection speed acquisition unit, or it can be obtained by differential calculation from the injection rod displacement signal; the injection pressure sensor can be installed at the hydraulic circuit of the injection cylinder or the pressure detection position of the pressure chamber to acquire the injection pressure.

[0038] The cavity status acquisition unit 12 is used to acquire the cavity pressure, vacuum pressure, and boost pressure of the current mold cycle and generate cavity response data. The cavity pressure sensor can be set near the gate, the end filling area, or the thin-walled area; the vacuum pressure sensor can be set in the vacuum pipeline or the cavity venting channel; and the boost pressure sensor can be set in the boost hydraulic circuit.

[0039] The mold thermal state acquisition unit 13 is used to acquire the mold zone temperature for the current mold cycle and generate zone thermal state data. Mold temperature sensors can be installed in the gate area, runner area, thin-walled area, thick-walled area, rib area, overflow area, and ejection area. The infrared thermal imaging acquisition unit can acquire mold surface temperature images after mold opening and convert the average temperature of the areas in the image into mold zone temperatures. The injection state acquisition module 1 may also include a demolding force sensor and a spraying state feedback unit for acquiring demolding force data and spraying state feedback data. The demolding force data is time-aligned and normalized by the data preprocessing module 2 and then incorporated into the current mold cycle process data to help determine the sticking trend in the ejection area.

[0040] The mold number identification unit 14 is connected to the injection curve acquisition unit 11, the cavity status acquisition unit 12, and the mold thermal status acquisition unit 13, respectively. It is used to associate the injection curve data, cavity response data, and zone thermal status data according to the current mold number to generate injection status data with mold number identification. The mold number identification can be triggered by the injection start signal, mold opening and closing signal, spraying completion signal, or die casting machine cycle counting signal.

[0041] In one embodiment, the sampling frequency for injection pressure, injection speed, injection rod displacement, and cavity pressure is 500Hz to 5000Hz; the sampling frequency for mold temperature is 1Hz to 50Hz; the infrared thermal imaging acquisition unit acquires mold surface temperature images within 0.5s to 5s after mold opening; and the sampling frequency for the demolding force sensor is 50Hz to 1000Hz. Through the above data acquisition methods, the flow state, pressure state, thermal state, and demolding state during the magnesium alloy die-casting process can be obtained simultaneously.

[0042] 2. Data preprocessing module 2; The data preprocessing module 2 includes a time synchronization unit 21, a working condition segmentation unit 22, a region mapping unit 23, and a normalized reference generation unit 24.

[0043] The time synchronization unit 21 is connected to the mold number identification unit 14 and is used to time-align the injection status data with the mold number identification and the pre-stored historical data based on the injection rod displacement characteristic point, the cavity pressure rise point, or the boost pressure rise point to obtain synchronized status data. The cavity pressure rise point can be defined as the moment when the cavity pressure exceeds the standard cavity pressure peak value by 5% to 15%; the boost pressure rise point can be defined as the moment when the boost pressure exceeds the standard boost pressure peak value by 5% to 20%.

[0044] The working condition segmentation unit 22 is connected to the time synchronization unit 21. It is used to divide the synchronization state data into a low-speed injection segment, a rapid filling segment, and a pressurized solidification segment based on the characteristics of injection speed and cavity pressure changes, thus obtaining segmented state data. The low-speed injection segment characterizes the propulsion process of the magnesium alloy melt within the pressure chamber; the rapid filling segment characterizes the filling process after the melt enters the cavity; and the pressurized solidification segment characterizes the solidification feeding and pressure holding processes.

[0045] The region mapping unit 23 is connected to the working condition segmentation unit 22. It is used to map the mold zone temperature in the segmented state data to each target mold region, and associate the cavity pressure arrival signal determined by the cavity response data with the corresponding target mold region to obtain regionalized process data. The target mold region includes at least one of the following: gate region, runner region, thin-walled region, thick-walled region, rib region, overflow region, and ejection region.

[0046] The normalized baseline generation unit 24 is connected to the region mapping unit 23. It is used to generate the current mold process data, historical thermal memory values, standard arrival time, and standard process data based on the regionalized process data and pre-stored historical data. The current mold process data and standard arrival time are output to the melt front identification module 3, and the historical thermal memory values ​​are output to the mold thermal memory generation module 4. The pre-stored historical data may include the injection curves, cavity pressure curves, mold zone temperatures, defect detection results, thermal memory values, and process parameter records of historical qualified molds.

[0047] To avoid errors caused by directly fusing data of different dimensions, data preprocessing module 2 uses a nonlinear saturated normalization function: ; in, For the first The normalization deviation value of the class variable ranges from 0 to 1; The deviation to be normalized; To and Permissible deviation scales with the same dimensions; The dimensionless stability coefficient can be taken as... ; This is a nonlinear sensitivity coefficient, with values ​​ranging from 1.2 to 2.5. Because... and Having the same dimensions, the input to the exponential function is a dimensionless quantity, and the formula dimensions match. (*) denotes an exponential function with the natural constant e as the base.

[0048] In this embodiment, , , , , , and All are nonlinear saturated normalization functions Specific application forms.

[0049] The function's purpose is to maintain a high resolution when there is a small deviation between the current module data and the standard data; when the deviation is large, the output gradually becomes saturated, thus avoiding excessive influence of a single abnormal peak on the melt front disturbance index, thermal memory matrix, and flow-thermal coupling mismatch index.

[0050] 3. Melt front identification module 3; The melt front identification module 3 is used to identify the actual flow state of magnesium alloy melt in the pressure chamber and mold cavity. The melt front identification module 3 includes a front signal extraction unit 31, an arrival time determination unit 32, a disturbance feature generation unit 33, a disturbance index generation unit 34, and a disturbance type determination unit 35.

[0051] The leading edge signal extraction unit 31 is connected to the data preprocessing module 2 and is used to extract the injection pressure, injection velocity, cavity pressure, and melt leading edge sensing signal formed by the injection state acquisition module 1 and the data preprocessing module 2 from the current mold process data. The melt leading edge sensing signal includes at least one of the cavity pressure arrival signal, ultrasonic echo signal, and electromagnetic induction signal. When ultrasonic echo or electromagnetic induction detection is not configured, the cavity pressure arrival signal is used as the basic melt leading edge sensing signal.

[0052] The arrival time determination unit 32 is connected to the leading edge signal extraction unit 31 and is used to determine the actual arrival time of the melt leading edge to each target mold area based on the cavity pressure rise point or the melt leading edge sensing signal change point corresponding to each target mold area. The cavity pressure rise point can be the moment when the cavity pressure exceeds the cavity pressure rise threshold and continues to exceed the pressure holding time threshold; the pressure holding time threshold can be 1ms to 10ms; the echo fluctuation threshold and electromagnetic fluctuation threshold can be determined based on the difference between the empty mold reference signal and the stable filling signal.

[0053] The disturbance feature generation unit 33 is connected to the arrival time determination unit 32, and is used to generate leading edge disturbance feature data based on the injection pressure change rate, injection velocity change rate, deviation between the actual arrival time and the standard arrival time, and the amplitude of the melt leading edge sensing signal fluctuation. The injection pressure change rate can be obtained by the ratio of the difference in injection pressure at adjacent sampling times to the sampling time interval, and the injection velocity change rate can be obtained by the ratio of the difference in injection velocity at adjacent sampling times to the sampling time interval. Before calculating the change rate, the injection pressure curve and the injection velocity curve can be subjected to moving average or low-pass filtering to reduce the influence of noise on the melt leading edge disturbance index.

[0054] The perturbation index generation unit 34 is connected to the perturbation feature generation unit 33, and is used to normalize and weightedly fuse the leading-edge perturbation feature data to generate the melt leading-edge perturbation index. The dynamic weights are: ; The melt front disturbance index is: ; in, The subscript for summing the perturbation features, =1,2,3,4 correspond to the injection pressure change rate, injection speed change rate, cavity pressure arrival time deviation, and melt front sensing signal fluctuation amplitude, respectively. For the first The availability flag for perturbation features takes a value of 0 or 1; when the first... When the sensing signal corresponding to the perturbation feature is available, =1, when the corresponding sensor signal is not configured, sampling is interrupted, or the effective range is exceeded. =0; For the first Reliability adjustment coefficient corresponding to disturbance-like characteristics; For the first The reliability of the sensor signal ranges from 0 to 1. Represents a system of terms with the natural constant e as the base and the constant e as the base. The value of the exponential function of the exponent; Represents the Sigmoid function; r=1,2,3,4 correspond to the injection pressure change rate, injection speed change rate, cavity pressure arrival time deviation, and melt front sensing signal fluctuation amplitude, respectively. , where is the dynamic weight of the r-th type of perturbation feature; This is the availability identifier for the r-th type of perturbation feature, with a value of 0 or 1; The value represents the reliability of the r-th type of sensor signal, ranging from 0 to 1. A larger value indicates a higher reliability of the corresponding sensor signal. This is the reliability adjustment factor; The dimensionless stability coefficient can be taken as... ; The rate of change of injection pressure; The rate of change of injection velocity; This represents the deviation between the actual arrival time and the standard arrival time of the cavity pressure. The amplitude of the ultrasonic echo signal or electromagnetic induction signal. and These are dimensionless model parameters; This represents the melt front disturbance index. Used to normalize the rate of change of injection pressure. Used to normalize the rate of change of injection velocity Used to normalize the deviation in cavity pressure arrival time. This is used to normalize the amplitude of the melt front sensing signal fluctuation. The allowable deviation scale of each of the above normalization functions is determined by the allowable deviation range of the corresponding variable in the standard process data.

[0055] When at least one perturbation feature is available Taking the minimum value is only used to improve numerical stability; it can also be omitted after confirming that the denominator is not zero. This ensures that the sum of all dynamic weights equals 1. When all perturbation features are unavailable, dynamic weights are not calculated, and the current module is marked as an invalid data module.

[0056] When the corresponding sensor signal is available =1; when the corresponding sensor signal is not configured, sampling is interrupted, or the signal exceeds the effective range. =0. The availability flags corresponding to the injection pressure change rate, injection velocity change rate, and cavity pressure arrival time deviation are typically set to 1. Without an ultrasonic echo sensor or electromagnetic induction sensor, the... Even with the corresponding availability flag set to 0, the system can still generate the melt front disturbance index based on the injection pressure change rate, injection speed change rate, and cavity pressure arrival time deviation.

[0057] parameter , and It can be stored in standard process data or pre-stored historical data, and can be updated by historical qualified module data and defect detection results. When historical samples are missing... Initialize to -2.0 to -1.0. Initialize to 1, Initialize to 0.1 to 1.0, and limit the parameter value range to -10 to 10.

[0058] This formula corresponds to melt front identification module 3. Its function is to: uniformly convert pressure surges, velocity fluctuations, arrival time deviations, and optional front sensing signals into a melt front disturbance index; dynamic weighting ensures that low-reliability or unavailable signals do not mislead the identification results; and interactive terms... It is used to enhance the precursor of the complex defect "abnormal pressure change and abnormal arrival time", which is helpful in identifying leading edge lag, leading edge splitting, air entrainment disturbance, backflow disturbance and impact filling.

[0059] The disturbance type determination unit 35 is connected to the disturbance index generation unit 34. It is used to determine the disturbance type of the melt front based on the comparison result between the melt front disturbance index and the preset disturbance threshold, and in combination with the actual arrival time difference between each target mold area, the vacuum pressure fluctuation characteristics, and the continuity of cavity pressure rise. The preset disturbance thresholds include 0.35, 0.60, and 0.80. A melt front is considered stable when the value is <0.35; 0.35≤ A value <0.60 indicates a slight disturbance; 0.60 ≤ A value <0.80 indicates a moderate disturbance; A value ≥0.80 is considered a severe disturbance.

[0060] The arrival time difference threshold can be 5ms to 30ms; the vacuum pressure fluctuation threshold can be 1.5 to 3 times the standard vacuum pressure fluctuation amplitude; and the pressure shock threshold can be 1.1 to 1.5 times the standard pressure peak value.

[0061] The types of melt front disturbances include at least one of the following: front lag, front splitting, air entrapment disturbance, backflow disturbance, and impact filling. Front lag corresponds to the melt arriving at the target mold area later than the standard arrival time; front splitting corresponds to the actual arrival time difference between different target mold areas exceeding the arrival time difference threshold; air entrapment disturbance corresponds to an abnormally high injection pressure change rate accompanied by vacuum pressure fluctuations; backflow disturbance corresponds to abnormal fluctuations in the injection speed curve and discontinuous rise in cavity pressure; impact filling corresponds to the pressure impact peak exceeding the pressure impact threshold in the second-speed stage. The melt front identification module 3 outputs the melt front disturbance index, the melt front disturbance type, and the actual arrival time to the fluid-thermal coupling mismatch judgment module 5.

[0062] 4. Mold thermal memory generation module 4; The mold thermal memory generation module 4 is used to identify the thermal accumulation state of the mold during continuous multi-mold production. Mold thermal memory differs from single mold temperature detection; it reflects the cumulative temperature trend of the mold area across multiple consecutive mold cycles.

[0063] The mold thermal memory generation module 4 includes a region temperature extraction unit 41, a historical thermal memory recall unit 42, a thermal memory recursion unit 43, a thermal state determination unit 44, and a thermal memory matrix output unit 45.

[0064] The area temperature extraction unit 41 is connected to the data preprocessing module 2 and is used to extract the mold zone temperature corresponding to each target mold zone from the current mold process data to obtain the current mold zone temperature data. The mold zone temperature matrix of the nth mold is represented as: ; in, Let n be the temperature matrix of the mold partitions for the nth mold. The actual temperature of the z-th target mold region in the n-th mold is given by ; Z represents the total number of target mold regions.

[0065] The historical thermal memory retrieval unit 42 is connected to the data preprocessing module 2. It is used to retrieve historical thermal memory values ​​and retrieve the standard thermal memory values, standard mold temperatures, cold boundary temperatures, thermal accumulation boundary temperatures, and thermal memory decay coefficients corresponding to each target mold area from the standard process data to obtain thermal memory reference data.

[0066] The thermal memory recursion unit 43 is connected to the region temperature extraction unit 41 and the historical thermal memory retrieval unit 42, respectively. It is used to recursively calculate the thermal accumulation state of each target mold region based on the current mold partition temperature data, historical thermal memory values, and thermal memory decay coefficients, generating a thermal memory matrix. To enable the thermal memory values ​​to distinguish between cold, thermally stable, and thermally accumulated states, this embodiment employs a directional temperature state function and a gated recursive thermal memory function. ; in, The dynamic thermal memory attenuation coefficient of the z-th target mold region in the n-th mold; ; This refers to the temperature of the region corresponding to the previous module. This refers to the production cycle time deviation between adjacent batches. Let Z be the dimensionless thermal sensitivity coefficient of the z-th target mold region; This represents the directional temperature state value of the z-th target mold region; The temperature of the relatively cool boundary of the z-th target mold region; Let be the thermal accumulation boundary temperature of the z-th target mold region, and > ; This is the current module's thermal memory value; This is the historical hot memory value corresponding to the previous module. to For dimensionless model parameters, Used to normalize the temperature changes between adjacent modules. Used to normalize the production cycle deviation between adjacent batches.

[0067] to Stored in standard process data or pre-stored historical data, it can be fitted based on historical qualified mold data, historical defective mold data, and trial mold data. When historical samples are missing... It can be initialized to -2.0 to -1.0. to It can be initialized to 0.1 to 1.0, and the parameter value range is limited to -10 to 10. All of the above parameters are dimensionless and are used to adjust the influence of temperature change, production cycle deviation, and regional thermal sensitivity coefficient on the dynamic thermal memory decay coefficient.

[0068] The directional temperature state function truncates the temperature of regions below the cooling boundary temperature to 0, truncates the temperature of regions above the heat accumulation boundary temperature to 1, and linearly maps the temperature of regions between the cooling boundary temperature and the heat accumulation boundary temperature to a range of 0 to 1. The standard mold temperature of the z-th target mold region is denoted as... The cooling boundary temperature and the heat accumulation boundary temperature can be determined based on the standard mold temperature. ; in, and To account for the temperature boundary offset of the target mold area, a range of 20°C to 40°C can be used, while ensuring... > .

[0069] When a previous second-heat memory value exists. Retrieved from pre-stored historical data; in the first valid module or when there is no previous module hot memory value, the formula... From the initial thermal memory value Replacement. Initial hot memory value Possible directional temperature state values Standard thermal memory value Or 0.50.

[0070] A fixed thermal memory decay coefficient λ can be used as A simplified form; when dynamic gating is not enabled, it makes =λ, where λ ranges from 0.5 to 0.9.

[0071] This formula corresponds to module 4 of the mold thermal memory generation module. Its function is to distinguish between undercooling and heat accumulation through a directional temperature state function, and to retain the thermal state influence of consecutive mold cycles through a gated recursive method, enabling the system to identify thermal drift and heat accumulation trends that are difficult to detect in a single temperature detection.

[0072] The thermal state determination unit 44 is connected to the thermal memory recursion unit 43, and is used to generate thermal state data for each target mold area based on the deviation of each thermal memory value from the standard thermal memory value, and in combination with a preset undercooling threshold, a preset thermal stability threshold, and a preset thermal accumulation threshold. The thermal state thresholds include an undercooling threshold of 0.30, a thermal stability threshold of 0.30 to 0.70, and a thermal accumulation threshold of 0.70. If the value is less than 0.30, the target mold area is considered too cold; if the value is less than or equal to 0.30, the target mold area is considered too cold. When the value is ≤0.70, the target mold area is determined to be in a thermally stable state; When the value is greater than 0.70, the target mold area is determined to be in a state of heat accumulation.

[0073] The thermal memory matrix output unit 45 is connected to the thermal state determination unit 44 and is used to output the thermal memory matrix and thermal state data to the thermal coupling mismatch determination module 5.

[0074] 5. Thermal-fluid coupling mismatch judgment module 5; The flow-thermal coupling mismatch judgment module 5 is used to determine whether the thermal state of a target mold area is suitable for melt filling and solidification when the front edge of the magnesium alloy melt reaches that area. The flow-thermal coupling mismatch judgment module 5 includes a region index association unit 51, a temperature gradient generation unit 52, a mismatch feature generation unit 53, a mismatch index generation unit 54, a mismatch level determination unit 55, a mismatch cause classification unit 56, and a mismatch result output unit 57.

[0075] The region index association unit 51 is connected to the melt front identification module 3 and the mold thermal memory generation module 4 respectively. It is used to associate the actual arrival time of the melt front to each target mold area, the melt front disturbance index and the thermal memory value of the corresponding target mold area in the thermal memory matrix with each target mold area as an index to obtain the region flow thermal association data.

[0076] The temperature gradient generation unit 52 is connected to the data preprocessing module 2 and the region index association unit 51. It is used to generate a normalized temperature gradient for each target mold region based on the mold partition temperature in the current mold process data. The normalized temperature gradient can be calculated based on the temperature difference between adjacent target mold regions and the allowable deviation scale of the standard temperature gradient.

[0077] The mismatch feature generation unit 53 connects the region index association unit 51 and the temperature gradient generation unit 52, and is used to generate region mismatch feature data based on the region flow heat association data, the deviation between the actual arrival time and the standard arrival time, the deviation between the thermal memory value and the standard thermal memory value, and the normalized temperature gradient.

[0078] The mismatch index generation unit 54 is connected to the mismatch feature generation unit 53. It is used to perform weighted fusion of regional mismatch feature data to generate the thermal-fluid coupling mismatch index corresponding to each target mold region, and then outputs the thermal-fluid coupling mismatch index to the defect risk vector generation module 6. The thermal-fluid coupling mismatch index is generated as follows: ; in, The thermal memory deviation of the z-th target mold region; This represents the standard thermal memory value for the z-th target mold region. The normalized deviation of the arrival time of the z-th target mold region; This represents the actual arrival time of the melt front reaching the z-th target mold region. Standard arrival time; The normalized temperature gradient for the z-th target mold region; to These are dimensionless model parameters; This represents the thermal-fluid coupling mismatch index. Because... , , , Both the interactive terms and the formulas are dimensionless, and the formulas match the dimensions. Used to normalize the melt front arrival time deviation of the z-th target mold region.

[0079] When historical samples are lacking, Initialize to -2.0 to -1.0. to Initialize to 0.1 to 1.0, and limit the parameter value range to -10 to 10.

[0080] This formula corresponds to module 5, which determines the flow-thermal coupling mismatch. Its function is to: [interaction item] Reflects the defect amplification effect when melt disturbance and regional thermal deviation are superimposed; interaction term This reflects the solidification mismatch caused by both arrival delay and local temperature gradient. In this way, the system can determine whether the thermal state of a region is suitable for filling and solidification when the melt arrives, thereby improving the accuracy of identifying precursors to defects such as cold shuts, air entrapment, shrinkage cavities, thermal cracks, and sticking.

[0081] The mismatch level determination unit 55 is connected to the mismatch index generation unit 54, and is used to compare the flow-thermal coupling mismatch index corresponding to each target mold area with a preset flow-thermal mismatch threshold to generate the flow-thermal mismatch level of each target mold area. The preset flow-thermal mismatch thresholds include 0.35, 0.60 and 0.80. A value <0.35 indicates normal heat flow matching; 0.35≤ A mismatch of <0.60 is considered mild; 0.60 ≤ A score <0.80 indicates a moderate mismatch; A value ≥0.80 indicates a severe mismatch.

[0082] The mismatch cause classification unit 56 is connected to the mismatch level determination unit 55, and is used to generate flow-thermal mismatch cause data based on the flow-thermal mismatch level, melt front disturbance index, deviation between actual arrival time and standard arrival time, deviation between thermal memory value and standard thermal memory value, normalized temperature gradient, and cavity pressure change characteristics extracted from the current mold process data. The flow-thermal mismatch cause data includes at least one of cold-state hysteresis mismatch, front disturbance mismatch, arrival time mismatch, thermal accumulation mismatch, and solidification pressure mismatch.

[0083] To enhance the ability to identify the causes of fluid-thermal mismatch, the fluid-thermal coupling mismatch judgment module 5 adopts the Softmax cause discrimination function: ; in, Let be the probability that the z-th target mold region belongs to the u-th type of flow thermal mismatch cause; u=1,2,3,4,5; u=1,2,3,4,5; u=1,2,3,4,5 correspond to cold state hysteresis mismatch, leading edge disturbance mismatch, arrival time mismatch, thermal accumulation mismatch and solidification pressure mismatch, respectively; Let z be the mismatch feature vector of the z-th target mold region; This is the normalized value of the cavity pressure variation characteristic corresponding to the z-th target mold region; Let be the dimensionless discriminant coefficient vector corresponding to the type u flow thermal mismatch cause. The superscript T denotes vector transpose. This represents the inner product between the discriminant coefficient vector corresponding to the type u flow thermal mismatch and the mismatch feature vector of the z-th target mold region. Since... Each component in the equation is dimensionless, and It is a dimensionless discriminant coefficient vector, therefore It is a dimensionless quantity.

[0084] Discriminant coefficient The data is stored in pre-stored mechanism-related data, which can be obtained from trial mold data, historical qualified molds and historical defective molds, or preset based on process experience.

[0085] when ≥0.60 and When <0.30, a cold-state hysteresis mismatch is determined; when ≥0.60 and When the value is greater than 0.50, an arrival time mismatch is determined; when... >0.70 and When the value is greater than 0.60, a thermal accumulation mismatch is determined to exist; when When the value is less than 0.30 and the normalized value of the cavity pressure rise slope is less than 0.50, it is determined that there is a risk of insufficient filling; when the thermal memory value of the ejection zone is greater than 0.70 and the peak value of the demolding force increases by more than 8% for three consecutive molds, it is determined that there is a tendency of sticking to the mold.

[0086] The mismatch result output unit 57 is connected to the mismatch cause classification unit 56, and is used to output the flow-heat coupling mismatch index, flow-heat mismatch level and flow-heat mismatch cause data to the defect risk vector generation module 6, so that the defect risk vector generation module 6 can generate the corresponding defect risk vector based on the flow-heat mismatch cause data.

[0087] The working principle of this embodiment is as follows: The injection state acquisition module 1 generates injection curve data, cavity response data, and zoned thermal state data. The data preprocessing module 2 transforms the multi-source asynchronous data into normalized process data for the target mold area. The melt front identification module 3 generates the melt front disturbance index, melt front disturbance type, and actual arrival time. The mold thermal memory generation module 4 generates a thermal memory matrix with cold and hot direction meanings. The flow-thermal coupling mismatch judgment module 5 outputs the flow-thermal coupling mismatch index, flow-thermal mismatch level, and flow-thermal mismatch cause data through a nonlinear flow-thermal coupling function and a causal discrimination function. Through this process, the system can couple and judge the melt flow state and the continuous thermal state of the mold, providing reliable input for subsequent defect risk vector generation and parameter contribution inversion.

[0088] Example 3, as Figure 7-9 As shown, this embodiment, based on Embodiments 1 and 2, further explains the specific structure, algorithm implementation, and control execution method of the defect risk vector generation module 6, parameter contribution inversion module 7, collaborative control execution module 8, and data feedback update module. This embodiment is used to realize defect risk vector generation, parameter contribution inversion, collaborative control execution, and data feedback update, solving the problems of difficulty in tracing the dominant process parameters of defect risks, difficulty in selecting the control object when multiple defect risks coexist, the possibility of introducing other defects through single parameter adjustment, and difficulty in adaptively updating the control effect in continuous production.

[0089] In this embodiment, i represents only the defect risk category number, j represents only the process parameter number, z represents only the target mold area number, Z represents the total number of target mold areas, and a and ℓ are used only as summation subscripts. The superscript T in the formula represents vector transpose, in the form of... The expression represents the inner product between two dimensionless vectors. Each risk component, regional feature, parameter contribution, and execution effectiveness is a dimensionless value, and the inputs to the exponential and sigmoid functions are also dimensionless values. The sigmoid function uses the definition from Example 2. .

[0090] 1. Defect Risk Vector Generation Module 6; The injection state acquisition module 1 is also used to acquire the demolding force data and spraying state feedback data of the current mold cycle. The data preprocessing module 2 is also used to perform time alignment and normalization processing on the demolding force data and spraying state feedback data, so that the processed demolding force data and spraying state feedback data are incorporated into the current mold cycle process data.

[0091] The defect risk vector generation module 6 is used to generate a defect risk vector based on the melt front disturbance index, thermal memory matrix, flow-thermal coupling mismatch index, flow-thermal mismatch level, and flow-thermal mismatch cause data. The defect risk vector generation module 6 includes a mismatch information receiving unit 61, a region defect mapping unit 62, a risk component generation unit 63, a risk level determination unit 64, and a risk vector output unit 65.

[0092] The mismatch information receiving unit 61 is connected to the thermal coupling mismatch judgment module 5. It receives the thermal coupling mismatch index corresponding to each target mold area and extracts vacuum pressure fluctuation characteristics, solidification stage cavity pressure change characteristics, mold zone temperature change characteristics, demolding force change characteristics, and spraying status feedback characteristics from the current mold process data to obtain basic data for defect judgment. Vacuum pressure fluctuation characteristics are used to judge venting stability; solidification stage cavity pressure change characteristics are used to judge the feeding and solidification pressure maintenance status; mold zone temperature change characteristics are used to judge local thermal field anomalies; demolding force change characteristics are used to judge ejection resistance and sticking tendency; and spraying status feedback characteristics are used to judge whether the release agent coverage, spraying time, and spraying position meet the process requirements.

[0093] The region defect mapping unit 62 is connected to the mismatch information receiving unit 61. It is used to associate each target mold region with at least one risk type among cold shut risk, air entrapment risk, oxidation inclusion risk, shrinkage cavity or porosity risk, hot cracking risk, and mold sticking risk, based on the location attributes and thermal flow coupling mismatch index of each target mold region, thus obtaining region defect mapping data. For example, thin-walled areas and end-filling areas are mainly associated with cold shut risk, undercasting risk, and air entrapment risk; gate areas and runner areas are mainly associated with air entrapment and oxidation inclusion risk; thick-walled areas are mainly associated with shrinkage cavity or porosity risk; rib areas with large temperature gradients and thick-thin interface areas are mainly associated with hot cracking risk; and ejection areas are mainly associated with mold sticking risk.

[0094] The risk component generation unit 63 is connected to the regional defect mapping unit 62 and is used to generate cold shut risk component, air entrapment risk component, oxidation inclusion risk component, shrinkage cavity or shrinkage porosity risk component, hot crack risk component and mold sticking risk component respectively based on regional defect mapping data, flow-heat coupling mismatch index and defect judgment basic data.

[0095] The defect risk vector is represented as: ; in, This represents the defect risk vector. Risk of cold insulation; Risk of vortexing; Risk of oxidation inclusions; Risk of shrinkage or porosity; Risk of thermal cracking; Risk of mold sticking; For the first Defect risk components. Each risk value ranges from 0 to 1.

[0096] To improve the causal targeting of defect risk generation, this embodiment employs a region attention risk function: ; ; in, For the first The regional attention weight of the defect risk for the z-th target mold area; Subscript for summation of regions; The dimensionless region feature vector is composed of the positional attributes, thermal memory value, temperature gradient, and thermal-fluid coupling mismatch index of the z-th target mold region. Indicates the first The dimensionless region feature vector, composed of the location attributes, thermal memory value, temperature gradient, and thermal-fluid coupling mismatch index of each target mold region, has a data structure similar to... same; For the first The dimensionless region attention coefficient vector corresponding to class defect risk; Indicates the first The dimensionless region attention coefficient vector of each target mold region The thermal-fluid coupling mismatch index for the z-th target mold region; The melt front disturbance index; This is a normalized value representing the vacuum pressure fluctuation characteristics. This represents the normalized value of the cavity pressure variation characteristics during the solidification stage; This is a normalized value representing the characteristic of the release force variation. This represents the normalized value of the spraying status feedback feature; to The risk fusion coefficient is a dimensionless coefficient. For the first Class defect risk component.

[0097] The superscript T indicates vector transpose. Indicates the first The inner product between the region attention coefficient vector corresponding to the defect risk and the region feature vector of the z-th target mold region. Because Let be a dimensionless region attention coefficient vector, and All features are dimensionless values, therefore It is a dimensionless quantity.

[0098] Regional attention coefficients and risk fusion coefficients are stored in pre-stored mechanism-related data and can be adjusted based on trial data, historical detection results, and data feedback update results. When historical samples are missing... Initialize to -2.0 to -1.0. to Initialize to 0.1 to 1.0, and limit the parameter value range to -10 to 10.

[0099] It is obtained by normalizing the amplitude of vacuum pressure fluctuations; It is obtained by normalizing the slope of the pressure drop in the cavity during the solidification stage or the characteristics of the sudden pressure drop. It is obtained by normalizing the rate of change of peak demolding force; It is obtained by normalizing the spraying flow rate deviation, spraying time deviation, or spraying position deviation.

[0100] This formula corresponds to Defect Risk Vector Generation Module 6. Its function is to automatically focus on target mold areas more relevant to their formation mechanism for different defect risks. For example, cold shut risk primarily focuses on thin-walled areas and end-filling areas, shrinkage cavity or porosity risk primarily focuses on thick-walled areas, and sticking risk primarily focuses on ejection areas. Through regional attention weighting, the system avoids treating all target mold areas equally, improving the causal orientation of defect risk judgment.

[0101] The defect risk assessment rules are used in conjunction with the regional attention risk function. When the thermal coupling mismatch index in the thin-walled region is greater than 0.60 and the thermal memory value in the thin-walled region is less than 0.30, the risk of cold shut increases. When the melt front disturbance index is greater than 0.60 and the normalized value of vacuum pressure fluctuation is greater than 0.50, the risk of gas entrapment increases. When the melt front disturbance index is greater than 0.60 and the melt exposure time exceeds the preset exposure time threshold, the risk of oxide inclusions increases. The preset exposure time threshold is 3s to 12s, which can be set according to the pouring distance and the magnesium alloy grade. When the thermal memory value in the thick-walled region is greater than 0.70 and the normalized value of the cavity pressure drop slope during solidification is greater than 0.50, the risk of shrinkage cavity or porosity increases. When the normalized value of the temperature gradient is greater than 0.60 and the cavity pressure exhibits a sudden drop during solidification, the risk of hot cracking increases. When the thermal memory value in the ejection zone is greater than 0.70 and the peak ejection force increases by more than 8% for three consecutive molds, the risk of sticking to the mold increases.

[0102] The aforementioned "improvement" refers to increasing the calculation input value, risk fusion coefficient corresponding item, or rule correction item of the corresponding defect risk component in the defect risk vector generation module 6, so that the corresponding risk component changes towards the high-risk range.

[0103] The risk level determination unit 64 is connected to the risk component generation unit 63, and is used to compare each risk component with its corresponding preset defect risk threshold to generate a defect risk level for each risk component. The preset defect risk thresholds include 0.35, 0.60, and 0.80. A risk value less than 0.35 is considered low risk; a risk value not less than 0.35 and not greater than 0.60 is considered medium risk; a risk value greater than 0.60 and not greater than 0.80 is considered high risk; and a risk value greater than 0.80 is considered severe risk.

[0104] The risk vector output unit 65 is connected to the risk level determination unit 64 and is used to combine the cold shut risk component, air entrapment risk component, oxidation inclusion risk component, shrinkage cavity or shrinkage porosity risk component, hot crack risk component and sticking risk component into a defect risk vector, and output the defect risk vector and defect risk level to the parameter contribution inversion module 7.

[0105] 2. Parameter contribution inversion module 7; The parameter contribution inversion module 7 is used to determine the dominant factors of defects based on the defect risk vector and to calculate the process parameters that should be adjusted first. The parameter contribution inversion module 7 includes a risk component receiving unit 71, a parameter deviation generation unit 72, a mechanism association calling unit 73, an execution effectiveness calling unit 74, a contribution calculation unit 75, a contribution ranking unit 76, a dominant parameter determination unit 77, a control direction generation unit 78, and an inversion result output unit 79.

[0106] The risk component receiving unit 71 is connected to the defect risk vector generation module 6 and is used to receive the defect risk vector and extract various defect risk components from it.

[0107] The parameter deviation generation unit 72 is connected to the data preprocessing module 2 and the collaborative control execution module 8, respectively. It extracts the current process parameters from the current module process data and the actuator status data from the previous or current module executed by the collaborative control execution module 8, and generates a process parameter deviation vector based on the difference between the current process parameters and the standard process data. The current process parameters include at least one of the following: low-speed injection speed, fast / slow speed switching point, second-speed speed, boost pressure, boost delay time, vacuum valve opening time, zone cooling intensity, and zone coating volume.

[0108] The target control parameter set is represented as follows: ; in, For the target set of control parameters; This refers to the low-speed injection velocity. This is the point where the speed switches between fast and slow. It is a second speed; For boosting pressure; This is the boost delay time; This refers to the moment the vacuum valve opens; For zoned cooling intensity; This refers to the amount of paint applied in each zone.

[0109] The mechanism correlation retrieval unit 73 is connected to the parameter deviation generation unit 72. It is used to retrieve the mechanism correlation coefficients between defect risks and process parameters from standard process data or pre-stored mechanism correlation data to form mechanism correlation data. The mechanism correlation coefficients can be preset based on process experience, historical data, and trial mold data, and can be updated during production. For example, cold shut risk is related to the second speed, thermal memory value, zone cooling intensity, and spraying amount; air entrapment risk is related to the low-speed injection speed, fast / slow speed switching point, and vacuum valve opening time; oxidation inclusion risk is related to melt exposure time, protective atmosphere, low-speed disturbance, and vacuum state; shrinkage cavity risk is related to the boost pressure, holding time, and cooling intensity of the thick-walled area; hot crack risk is related to the boost pressure, local temperature gradient, mold opening time, and cooling intensity; and mold sticking risk is related to the ejection zone thermal memory value, spraying amount, spraying position, and mold opening time.

[0110] The execution effectiveness retrieval unit 74 is connected to the data preprocessing module 2. It retrieves the adjustment effectiveness of each actuator on the corresponding process parameters from standard process data or pre-stored historical data, forming execution effectiveness data. Execution effectiveness characterizes the feasibility of the actuator effectively adjusting the process parameters. For example, if the cooling water proportional valve has a delayed response or the spraying robot's flow rate fluctuates significantly, the corresponding execution effectiveness decreases; if an actuator's adjustment can stably reduce the corresponding defect risk, the corresponding execution effectiveness increases.

[0111] The contribution calculation unit 75 is connected to the risk component receiving unit 71, the parameter deviation generation unit 72, the mechanism association calling unit 73, and the execution effectiveness calling unit 74, respectively. It is used to calculate the parameter contribution corresponding to each current process parameter based on the defect risk vector, process parameter deviation vector, mechanism association data, and execution effectiveness data. The parameter contribution inversion module 7 uses a normalized competitive contribution function. ; ; in, The original contribution score for the j-th process parameter. For the first The original contribution score of each target control parameter; The contribution of the normalized parameter to the j-th process parameter; For the first Class defect risk; For the first The dimensionless mechanistic correlation coefficient between the defect risk and the j-th process parameter; This represents the current value of the j-th process parameter; This is the standard value of the j-th process parameter; This is the normalized value of the deviation of the j-th process parameter; Let be the dimensionless regulation effectiveness of the j-th actuator; , and The value is a dimensionless adjustment coefficient, which can be 0.20, 0.10 and 0.15 respectively.

[0112] For nonlinear saturated normalization function The specific application form for the j-th process parameter deviation. Among them, The deviation between the current value and the standard value of the j-th process parameter is defined by the allowable adjustment range of the j-th process parameter in the standard process data.

[0113] The value ranges from 0 to 1, with a larger value indicating the first... The higher the degree of mechanistic correlation between the type of defect risk and the j-th process parameter; The value ranges from 0 to 1. The larger the value, the more reliable the adjustment of the corresponding process parameters by the j-th actuator. The value range is from 0 to 1, and all target control parameters correspond to... The sum is 1. Used to adjust the weight of the influence of process parameter deviation on the parameter contribution. Used to adjust the weight of the impact of execution effectiveness on the contribution of parameters. This is used to adjust the weight of the interaction term between "the degree of correlation between defect risk and process parameter mechanism" and "execution effectiveness" on the contribution of parameters.

[0114] This formula corresponds to Parameter Contribution Inversion Module 7. Its function is: It reflects the mechanistic correlation between defect risk and process parameters; This reflects the degree to which the current process parameters deviate from the standard process data; This reflects the feasibility of the actuator effectively adjusting the j-th process parameter; It reflects a comprehensive priority that prioritizes high-risk, highly correlated, and effectively executable parameters. Softmax normalization allows multiple candidate process parameters to compete on the same scale, facilitating the determination of priority control parameters.

[0115] The contribution ranking unit 76 is connected to the contribution calculation unit 75 and is used to rank the contribution of each parameter to obtain the parameter contribution ranking result.

[0116] The dominant parameter determination unit 77 is connected to the contribution ranking unit 76, and is used to determine at least one dominant process parameter based on the parameter contribution ranking result, and to determine the dominant process parameter as the priority control parameter. When the difference between the contribution of the highest normalized parameter and the contribution of the second highest normalized parameter is less than 0.10, the system simultaneously determines multiple corresponding process parameters as priority control parameters; when the contribution of the highest normalized parameter is greater than 0.70, the system generates a strong correction flag. When it is necessary to weaken the influence of execution effectiveness on the defect cause judgment, the system can first determine the priority control parameter based on the parameter contribution ranking result. and Determine candidate dominant parameters, and then utilize Sort among the candidate dominant parameters.

[0117] The control direction generation unit 78 is connected to the dominant parameter determination unit 77. It is used to generate control direction data for reducing the corresponding defect risk based on the defect risk component, process parameter deviation vector and mechanism correlation coefficient corresponding to the priority control parameter, and generate a strong correction mark when the parameter contribution is greater than 0.70.

[0118] The inversion result output unit 79 is connected to the control direction generation unit 78, and is used to output the priority control parameters, control direction data and strong correction flag to the collaborative control execution module 8.

[0119] 3. Coordinated Control and Execution Module 8; The collaborative control execution module 8 is used to generate control commands based on the parameter contribution inversion results, and output the control commands to actuators such as the injection controller, vacuum valve controller, mold temperature controller, cooling water proportional valve, spraying robot, and pressurization controller. The collaborative control execution module 8 includes a control strategy matching unit 81, a current mold fast closed-loop unit 82, a next mold slow closed-loop unit 83, a safety boundary verification unit 84, and a control command output unit 85.

[0120] The control strategy matching unit 81 is connected to the parameter contribution inversion module 7. It receives priority control parameters and control direction data, and determines the adjustable parameters for the current cycle and the correction parameters for the next cycle based on the defect risk component and control direction data corresponding to the priority control parameters. The adjustable parameters for the current cycle include low-speed injection speed, fast / slow speed switching point, second-speed speed, pressurization delay time, and vacuum valve opening time. The correction parameters for the next cycle include pressurization pressure, zone cooling intensity, zone coating volume, and initial injection curve.

[0121] The current mold rapid closed-loop unit 82 is connected to the control strategy matching unit 81, and is used to generate rapid closed-loop control commands based on the adjustable parameters of the current mold while the current mold is still within the preset adjustable time window. The preset adjustable time window is from the start of injection to the start of pressurization, or it can be dynamically determined based on the displacement of the injection rod and the starting point of the cavity pressure.

[0122] When the melt front disturbance index is greater than 0.60 and the fast / slow speed switching position has not yet been reached, the low-speed injection speed is reduced by 5% to 15%, and the fast / slow speed switching point is delayed by 1mm to 5mm; when the normalized deviation of the arrival time corresponding to the thin-walled region is greater than 0.50, and the thermal memory value of the thin-walled region is less than 0.30, the second-speed speed is increased by 3% to 10%, and the pressurization delay time is shortened by 5ms to 20ms; when the normalized value of vacuum pressure fluctuation is greater than 0.50 and the melt front disturbance index is greater than 0.60, the vacuum valve opening time is advanced by 20ms to 100ms, and the initial acceleration of the second-speed section is reduced by 5% to 20%.

[0123] The next-stage slow closed-loop unit 83 is connected to the control strategy matching unit 81 and is used to generate slow closed-loop control commands based on the next-stage correction parameters. The slow closed-loop control commands are used to correct at least one of the following parameters for the next stage: boost pressure, zone cooling intensity, zone spraying amount, and initial injection curve.

[0124] When the cold shut risk component in the thin-walled area is greater than 0.60, the cold shut risk in the thin-walled area is determined to be high. The coating amount in the thin-walled area is reduced by 5% to 20%, the cooling water flow rate in the thin-walled area is reduced by 5% to 15%, and the second speed of the next mold is increased. When the shrinkage cavity or porosity risk component in the thick-walled area is greater than 0.60, the shrinkage cavity risk in the thick-walled area is determined to be high. The cooling water flow rate in the thick-walled area is increased by 5% to 20%, the boost pressure is increased by 3% to 10%, and the pressure holding time is extended by 10ms to 50ms. When the ejector zone sticking risk component is greater than 0.60, the ejector zone sticking risk is considered high, and the ejector zone coating amount is increased by 5% to 25%, and the coating time is extended by 0.1s to 0.5s. When the ejector zone sticking risk component is greater than 0.60 and the ejector zone thermal memory value is greater than 0.70, the ejector zone cooling water valve opening is increased by 5% to 15%. When the ejector zone mold zone temperature reaches more than 90% of the upper limit of mold temperature, the cooling water valve opening is increased first. When the normalized value of the coating flow deviation, coating time deviation, or coating position deviation is greater than 0.50, the coating status feedback characteristics are considered abnormal, and the coating amount or coating time is increased first. When the air entrapment risk component is greater than 0.60, the air entrapment risk is considered high, the low-speed acceleration is reduced by 5% to 20%, the fast / slow speed switching point is advanced or delayed by 1mm to 5mm, and the vacuum valve opening time is advanced by 20ms to 100ms.

[0125] The safety boundary verification unit 84 is connected to the current mold's fast closed-loop unit 82 and the next mold's slow closed-loop unit 83, respectively. It compares the fast and slow closed-loop control commands with preset safety boundaries, and limits or replaces the corresponding control command with a suboptimal control command when the threshold value exceeds the corresponding safety boundary, thus obtaining a safe control command. The preset safety boundaries include the maximum injection speed boundary, the maximum boost pressure boundary, the upper limit of mold temperature, the upper limit of cooling water valve opening, the upper limit of coating amount, the vacuum valve opening time boundary, and the clamping force safety boundary. In one embodiment, the maximum injection speed boundary is 5.0 m / s, the maximum boost pressure boundary is 90 MPa, the upper limit of mold temperature is 260°C, the upper limit of cooling water valve opening is 100%, the upper limit of coating amount is 140% of the standard coating amount, the vacuum valve opening time boundary is 30 ms to 120 ms after injection start, and the clamping force safety boundary is 90% of the rated clamping force of the die-casting machine. When the control command exceeds any safety boundary, the safety boundary verification unit 84 limits the corresponding control amount; when the limit still cannot reduce the corresponding defect risk, the safety boundary verification unit 84 selects the suboptimal control parameter according to the parameter contribution ranking result.

[0126] The control command output unit 85 is connected to the safety boundary verification unit 84 and is used to output safety control commands to at least one of the following: injection controller, booster controller, vacuum valve controller, cooling controller or cooling water proportional valve, and spraying actuator.

[0127] 4. Data feedback and update module; The data feedback and update module is used to feed back die-casting process data, actuator feedback data, demolding force data, demolding thermal imaging data, and subsequent test results into the system to update standard process parameters, defect risk judgment rules, and parameter contribution weights.

[0128] Subsequent inspection results include appearance inspection, dimensional inspection, X-ray inspection, airtightness inspection, weight inspection, and mechanical property sampling inspection. When a defect is confirmed by subsequent inspection, the system increases the mechanism correlation coefficient corresponding to that defect; when the defect risk decreases significantly after the execution of a certain control parameter, the system increases the effectiveness of the execution of that control parameter; when the defect risk is not improved or other defects are introduced after the execution of a certain control parameter, the system decreases the effectiveness of the execution of that control parameter. Through the above methods, the system can adapt to mold wear, material batch changes, environmental temperature and humidity changes, and production cycle changes.

[0129] The working principle of this embodiment is as follows: The defect risk vector generation module 6 generates six types of defect risk components based on the thermal-fluid coupling mismatch index, regional attributes, and basic defect judgment data; the parameter contribution inversion module 7 determines priority control parameters based on defect risk, process parameter deviation, mechanism correlation coefficient, and execution effectiveness; the collaborative control execution module 8 generates safety control instructions based on the different adjustable windows of the current and next modules; and the data feedback update module updates subsequent judgment and control basis based on execution and detection results. Through this closed-loop process, the system can transform defect risk judgment into process adjustment actions with causal orientation, reduce fluctuations caused by blindly adjusting all parameters, reduce the risk of one defect being suppressed while another is amplified, and improve the stability of the magnesium alloy die-casting process and the consistency of casting quality.

[0130] The above are merely preferred embodiments of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. An online quality control system for magnesium alloy die casting process, characterized in that, include: The injection status acquisition module is used to acquire the injection status data of the current cycle. The data preprocessing module is connected to the injection state acquisition module. It is used to perform time alignment, segmentation and normalization on the injection state data and the pre-stored historical data to obtain the target mold area, the current mold process data, historical thermal memory values, standard arrival time and standard process data. The melt front identification module, connected to the data preprocessing module, is used to generate the melt front disturbance index and the actual arrival time of each target mold area based on the injection pressure change rate, injection speed change rate, and the deviation between the cavity pressure arrival time and the standard arrival time in the current mold process data. The mold thermal memory generation module, connected to the data preprocessing module, is used to generate a thermal memory matrix based on the mold zone temperature and historical thermal memory values. The fluid-thermal coupling mismatch judgment module connects the melt front identification module and the mold thermal memory generation module. It is used to calculate the fluid-thermal coupling mismatch index by coupling the melt front disturbance index, actual arrival time and thermal memory matrix with the target mold area as the index. The defect risk vector generation module, connected to the flow-thermal coupling mismatch judgment module, is used to generate defect risk vectors for multiple types of defect risks based on the flow-thermal coupling mismatch index. The parameter contribution inversion module connects the defect risk vector generation module and the data preprocessing module. It is used to determine the priority control parameters based on the defect risk vector, the deviation of the current process parameters from the standard process data, and the mechanism correlation coefficient between defect risk and process parameters. The collaborative control execution module, connected to the parameter contribution inversion module, is used to generate safety control instructions for the current and / or next mode within a preset safety boundary based on the priority control parameters, so as to adjust at least one of the injection, pressurization, vacuum, cooling, and spraying parameters.

2. The online quality control system for magnesium alloy die casting process according to claim 1, characterized in that, The injection status acquisition module includes: The injection curve acquisition unit is used to acquire the injection pressure, injection speed and injection rod displacement of the current module and generate injection curve data; The cavity status acquisition unit is used to acquire the cavity pressure, vacuum pressure, and boost pressure of the current mold and generate cavity response data; The mold thermal state acquisition unit is used to acquire the temperature of the mold zone in the current mold batch and generate zone thermal state data; The mold number identification unit is connected to the injection curve acquisition unit, the cavity status acquisition unit and the mold thermal status acquisition unit respectively. It is used to associate the injection curve data, the cavity response data and the partition thermal status data according to the current mold number to generate injection status data with mold number identification. The data preprocessing module includes: The time synchronization unit, connected to the mold identification unit, is used to time-align the injection status data with the mold identification and the pre-stored historical data based on the injection rod displacement characteristic point, the cavity pressure rise point, or the pressurization pressure rise point to obtain synchronized status data. The working condition segmentation unit, connected to the time synchronization unit, is used to divide the synchronization state data into a low-speed injection segment, a fast filling segment, and a pressurized solidification segment according to the characteristics of the injection speed change and the characteristics of the cavity pressure change, so as to obtain segmented state data. The region mapping unit, connected to the working condition segmentation unit, is used to map the mold partition temperature in the segmented state data to each of the target mold regions, and associate the cavity pressure arrival signal determined by the cavity response data with the corresponding target mold region to obtain regionalized process data; The normalized baseline generation unit, connected to the region mapping unit, is used to generate the current mold process data, the historical thermal memory value, the standard arrival time, and the standard process data based on the regionalized process data and the pre-stored historical data. The current mold process data and the standard arrival time are output to the melt front identification module, and the historical thermal memory value is output to the mold thermal memory generation module.

3. The online quality control system for magnesium alloy die casting process according to claim 1, characterized in that, The melt front identification module includes: The leading edge signal extraction unit, connected to the data preprocessing module, is used to extract from the current process data the injection pressure, injection speed, cavity pressure, and melt leading edge sensing signal formed by the injection state acquisition module and the data preprocessing module. The melt leading edge sensing signal includes at least one of cavity pressure arrival signal, ultrasonic echo signal, and electromagnetic induction signal. The arrival time determination unit is connected to the leading edge signal extraction unit and is used to determine the actual arrival time of the melt leading edge to each of the target mold areas based on the cavity pressure rise point or the melt leading edge sensing signal change point corresponding to each of the target mold areas. The disturbance feature generation unit, connected to the arrival time determination unit, is used to generate leading edge disturbance feature data based on the rate of change of the injection pressure, the rate of change of the injection velocity, the deviation between the actual arrival time and the standard arrival time, and the fluctuation amplitude of the melt leading edge sensing signal. The disturbance index generation unit, connected to the disturbance feature generation unit, is used to normalize and weighted fuse the leading edge disturbance feature data to generate the melt leading edge disturbance index; The disturbance type determination unit, connected to the disturbance index generation unit, is used to determine the type of melt front disturbance based on the comparison result of the melt front disturbance index and the preset disturbance threshold, combined with the actual arrival time difference between each target mold area, the vacuum pressure fluctuation characteristics extracted from the current mold process data, and the continuity of cavity pressure rise. The type of melt front disturbance includes at least one of front lag, front split, air entrapment disturbance, backflow disturbance, and impact filling. The unit outputs the melt front disturbance index, the type of melt front disturbance, and the actual arrival time to the flow-thermal coupling mismatch determination module.

4. The online quality control system for magnesium alloy die casting process according to claim 1, characterized in that, The mold thermal memory generation module includes: The area temperature extraction unit is connected to the data preprocessing module and is used to extract the mold partition temperature corresponding to each target mold area from the current mold process data to obtain the current mold partition temperature data. The historical thermal memory retrieval unit is connected to the data preprocessing module and is used to retrieve the historical thermal memory value and retrieve the standard thermal memory value and thermal memory decay coefficient corresponding to each target mold area from the standard process data to obtain thermal memory reference data. The thermal memory recursion unit is connected to the area temperature extraction unit and the historical thermal memory recall unit, respectively. It is used to recursively calculate the thermal accumulation state of each target mold area based on the current module partition temperature data, the historical thermal memory value and the thermal memory decay coefficient, and generate the thermal memory matrix. The thermal memory matrix includes the thermal memory value corresponding to each target mold area. A thermal state determination unit, connected to the thermal memory recursion unit, is used to generate thermal state data for each target mold area based on the deviation between each thermal memory value and the standard thermal memory value, and in combination with a preset cooling threshold, a preset thermal stability threshold, and a preset thermal accumulation threshold. The thermal memory matrix output unit is connected to the thermal state determination unit and is used to output the thermal memory matrix and the thermal state data to the thermal coupling mismatch determination module.

5. The online quality control system for magnesium alloy die casting process according to claim 1, characterized in that, The thermal-fluid coupling mismatch determination module includes: The region index association unit is connected to the melt front identification module and the mold thermal memory generation module respectively. It is used to associate the actual arrival time of the melt front to each target mold area, the melt front disturbance index, and the thermal memory value of the corresponding target mold area in the thermal memory matrix with each target mold area as an index to obtain region flow thermal association data. A temperature gradient generation unit, connected to the data preprocessing module and the region index association unit, is used to generate a normalized temperature gradient for each target mold region based on the mold partition temperature in the current mold process data. The mismatch feature generation unit, connected to the region index association unit and the temperature gradient generation unit, is used to generate region mismatch feature data based on the region flow-heat association data, the deviation between the actual arrival time and the standard arrival time, the deviation between the thermal memory value and the standard thermal memory value obtained from the standard process data, and the normalized temperature gradient. The mismatch index generation unit, connected to the mismatch feature generation unit, is used to perform weighted fusion of the regional mismatch feature data to generate the flow-thermal coupling mismatch index corresponding to each target mold region, and output the flow-thermal coupling mismatch index to the defect risk vector generation module.

6. The online quality control system for magnesium alloy die casting process according to claim 5, characterized in that, The thermal-fluid coupling mismatch determination module further includes: The mismatch level determination unit, connected to the mismatch index generation unit, is used to compare the flow-thermal coupling mismatch index corresponding to each of the target mold areas with a preset flow-thermal mismatch threshold to generate the flow-thermal mismatch level of each target mold area; The mismatch cause classification unit, connected to the mismatch level determination unit, is used to generate flow-thermal mismatch cause data based on the flow-thermal mismatch level, the melt front disturbance index, the deviation between the actual arrival time and the standard arrival time, the deviation between the thermal memory value and the standard thermal memory value, the normalized temperature gradient, and the cavity pressure change characteristics extracted from the current mold process data. The flow-thermal mismatch cause data includes at least one of cold-state hysteresis mismatch, front disturbance mismatch, arrival time mismatch, thermal accumulation mismatch, and solidification pressure mismatch. The mismatch result output unit is connected to the mismatch cause classification unit and is used to output the flow-heat coupling mismatch index, the flow-heat mismatch level and the flow-heat mismatch cause data to the defect risk vector generation module, so that the defect risk vector generation module generates a corresponding defect risk vector based on the flow-heat mismatch cause data.

7. The online quality control system for magnesium alloy die casting process according to claim 1, characterized in that, The injection state acquisition module is also used to acquire the demolding force data and spraying state feedback data of the current mold cycle. The data preprocessing module is also used to perform time alignment and normalization processing on the demolding force data and the spraying state feedback data, so that the processed demolding force data and spraying state feedback data are incorporated into the current mold cycle process data. The defect risk vector generation module includes: The mismatch information receiving unit is connected to the thermal coupling mismatch judgment module. It is used to receive the thermal coupling mismatch index corresponding to each target mold area, and extract the vacuum pressure fluctuation characteristics, solidification stage cavity pressure change characteristics, mold partition temperature change characteristics, demolding force change characteristics and spraying status feedback characteristics from the current mold process data to obtain the basic data for defect judgment. The regional defect mapping unit, connected to the mismatch information receiving unit, is used to associate the target mold area with at least one of the following risk types according to the location attributes of each target mold area and the flow-heat coupling mismatch index, thereby obtaining regional defect mapping data. The risk component generation unit, connected to the region defect mapping unit, is used to generate, respectively, cold shut risk component, air entrapment risk component, oxidation inclusion risk component, shrinkage cavity or shrinkage porosity risk component, hot crack risk component and mold sticking risk component based on the region defect mapping data, the flow-heat coupling mismatch index and the defect judgment basic data. The risk level determination unit is connected to the risk component generation unit and is used to compare each risk component with the corresponding preset defect risk threshold to generate the defect risk level corresponding to each risk component. The risk vector output unit, connected to the risk level determination unit, is used to combine the cold shut risk component, the air entrapment risk component, the oxidation inclusion risk component, the shrinkage cavity or shrinkage porosity risk component, the hot crack risk component, and the mold sticking risk component into the defect risk vector, and output the defect risk vector and the defect risk level to the parameter contribution inversion module.

8. The online quality control system for magnesium alloy die casting process according to claim 1, characterized in that, The parameter contribution inversion module includes: A risk component receiving unit, connected to the defect risk vector generation module, is used to receive the defect risk vector and extract various defect risk components therein. The parameter deviation generation unit is connected to the data preprocessing module and the collaborative control execution module, respectively. It is used to extract the current process parameters from the current module process data and the actuator status data of the previous or current module executed by the collaborative control execution module, and generate a process parameter deviation vector according to the difference between the current process parameters and the standard process data. The current process parameters include at least one of the following: low-speed injection speed, fast-slow speed switching point, second speed, boost pressure, boost delay time, vacuum valve opening time, zone cooling intensity, and zone spraying amount. The mechanism association calling unit is connected to the parameter deviation generation unit and is used to call the mechanism association coefficient between defect risk and process parameters from the standard process data or pre-stored mechanism association data to form mechanism association data; An execution effectiveness retrieval unit, connected to the data preprocessing module, is used to retrieve the adjustment effectiveness of each actuator on the corresponding process parameters from the standard process data or pre-stored historical data, forming execution effectiveness data; The contribution calculation unit is connected to the risk component receiving unit, the parameter deviation generation unit, the mechanism association calling unit, and the execution effectiveness calling unit, respectively, and is used to calculate the parameter contribution corresponding to each current process parameter based on the defect risk vector, the process parameter deviation vector, the mechanism association data, and the execution effectiveness data.

9. The online quality control system for magnesium alloy die casting process according to claim 8, characterized in that, The parameter contribution inversion module also includes: A contribution ranking unit, connected to the contribution calculation unit, is used to rank the contribution of each parameter to obtain a parameter contribution ranking result. A dominant parameter determination unit, connected to the contribution ranking unit, is used to determine at least one dominant process parameter based on the parameter contribution ranking result, and to determine the dominant process parameter as the priority control parameter. When the difference between the highest parameter contribution and the second highest parameter contribution is less than a preset contribution difference threshold, the corresponding multiple process parameters are determined as the priority control parameters. The control direction generation unit is connected to the dominant parameter determination unit. It is used to generate control direction data for reducing the corresponding defect risk based on the defect risk component corresponding to the priority control parameter, the process parameter deviation vector and the mechanism correlation coefficient, and generate a strong correction flag when the parameter contribution is greater than a preset strong correction threshold. The inversion result output unit is connected to the control direction generation unit and is used to output the priority control parameters, the control direction data and the strong correction identifier to the collaborative control execution module.

10. The online quality control system for magnesium alloy die casting process according to claim 9, characterized in that, The coordinated control execution module includes: The control strategy matching unit is connected to the parameter contribution inversion module and is used to receive the priority control parameters and the control direction data, and determine the adjustable parameters of the current module and the correction parameters of the next module based on the defect risk component corresponding to the priority control parameters and the control direction data. The current module rapid closed-loop unit is connected to the control strategy matching unit. When the current module is still within the preset adjustable time window, it generates a rapid closed-loop control command based on the adjustable parameters of the current module. The rapid closed-loop control command is used to adjust at least one of the following: low-speed injection speed, fast-slow speed switching point, second speed, pressurization delay time, and vacuum valve opening time. The next-mode slow closed-loop unit is connected to the control strategy matching unit and is used to generate a slow closed-loop control command based on the next-mode correction parameters. The slow closed-loop control command is used to correct at least one of the following in the next-mode: boost pressure, zone cooling intensity, zone spraying amount, and initial injection curve. A safety boundary verification unit is connected to the current module fast closed-loop unit and the next module slow closed-loop unit, respectively. It is used to compare the fast closed-loop control command and the slow closed-loop control command with the preset safety boundary, and limit or replace the corresponding control command with a suboptimal control command when the corresponding safety boundary threshold is exceeded, so as to obtain the safety control command. The control command output unit is connected to the safety boundary verification unit and is used to output the safety control command to at least one of the following: injection controller, booster controller, vacuum valve controller, cooling controller or cooling water proportional valve, and spraying actuator.

Images