Data-driven radiator production process parameter adaptive optimization method

Abstract

This invention discloses a data-driven adaptive optimization method for heat sink manufacturing process parameters, belonging to the field of machine learning. By acquiring the thermal input data of the target SoC chip under target operating conditions, the manufacturing process parameter data of the SoC heat sink in the current production batch, and production status detection data, heat transfer status data is generated based on the heat transfer relationship between the target SoC chip, the heat sink substrate, and the heat sink fins. The loading data of the initial thermal performance simulation model is corrected to obtain the target thermal performance simulation model. Simulation evaluation of different candidate manufacturing process parameter schemes determines the target manufacturing process parameter scheme. This helps solve the problem that existing manufacturing parameter schemes determined based on simulation evaluation results are difficult to improve contact thermal resistance, heat diffusion uniformity, and temperature rise control effects in the actual heat dissipation process, thereby reducing the accuracy and reliability of SoC heat sink manufacturing process parameter optimization results.

Description

Technical Field

[0001] This invention relates to the field of machine learning technology, and in particular to a data-driven adaptive optimization method for heat sink manufacturing process parameters. Background Technology

[0002] As an important component for heat transfer, heat diffusion and temperature control, radiators are widely used in automotive cooling, industrial equipment temperature control, communication equipment heat dissipation and electronic product thermal management. Their heat dissipation performance and structural reliability directly affect the operational stability and service life of the supporting equipment.

[0003] As electronic devices increasingly move towards higher integration, miniaturization, and higher computing performance, the heat generated by internal chips within limited installation space is constantly increasing. Traditional methods relying on natural heat dissipation from the casing or simple heat-dissipating pads are insufficient to meet the temperature control requirements of high-power chips during continuous operation. This is especially true for System-on-Chip (SoC), which integrates processors, memory control, communication interfaces, and functional computing units into a single chip or package structure. Under high load conditions, this can easily lead to localized heat concentration and excessive temperature rise. Therefore, a SoC heatsink corresponding to the SoC chip's package surface or thermal cover is required. This heatsink, through components such as a heat dissipation substrate, heat dissipation fins, heat diffusion structures, and thermal interface materials, effectively transfers and releases the heat generated by the SoC chip to the external environment. The contact fit of the SoC heatsink, the continuity of the heat conduction path, the precision of the heat dissipation structure, and the heat diffusion capability directly affect the operating temperature, operational reliability, and lifespan of the SoC chip.

[0004] Taking a finned heatsink used for surface heat dissipation of SoC chips as an example, its production and verification process typically includes steps such as preparation of thermally conductive metal materials, processing of the heat dissipation substrate, forming of heat dissipation fins, connection and forming of the heat dissipation substrate and heat dissipation fins, finishing of the chip contact bottom surface, surface cleaning and protective treatment, finished product dimensional inspection, and thermal performance verification. When applied to SoC chip heat dissipation components, thermally conductive interface material can be placed between the heatsink contact bottom surface and the SoC chip package surface or thermally conductive cover plate according to assembly requirements, and the heatsink can be positioned and pressed to form a heat conduction path for heat transfer from the SoC chip to the heatsink substrate and heatsink fins. Among them, the heatsink substrate is used to receive and diffuse the local heat generated by the SoC chip, the heatsink fins are used to increase the heat exchange area between the heatsink and the outside air, the finishing of the chip contact bottom surface is used to control the contact conditions between the heatsink and the corresponding contact area, the thermally conductive interface material is used to fill the tiny gaps on the contact surface and reduce the interface thermal resistance, and thermal resistance detection, temperature rise detection, and thermal diffusion performance detection are used to determine whether the heatsink meets the predetermined heat dissipation requirements under the corresponding SoC chip heat dissipation conditions.

[0005] To improve the accuracy of parameter settings and the stability of product heat dissipation performance during the production of SoC heat sinks, existing technologies typically record relevant data in various stages of SoC heat sink production and adaptation verification, based on the package size, power consumption level, heat sink structure specifications, thermal conductive material batch, and production batch of the SoC chip to be adapted. This includes data on heat sink substrate processing, heat sink fin forming, heat sink body connection, contact bottom surface treatment, thermal interface material configuration, and assembly bonding. Recordable structural and testing data include fin forming dimensions, fin spacing, contact bottom surface flatness, surface roughness, appearance dimensional inspection results, thermal resistance test results, chip simulated heat source temperature rise, temperature distribution uniformity, and heat dissipation reliability test results. By analyzing production records and test results, and using machine learning models, such as thermal performance prediction simulation models based on regression analysis, decision trees, support vector regression, or neural networks, the correlation between the heat sink structure state, thermal contact conditions, and changes in some processing parameters and the heat dissipation performance can be determined. Based on this, relevant parameters in subsequent production batches or assembly verification processes can be adjusted to reduce the probability of problems such as increased contact thermal resistance, excessive local temperature rise, or uneven heat diffusion. Among these parameters, the SoC heat sink manufacturing process parameters include, but are not limited to: the spindle speed, feed rate, and depth of cut when the heat sink substrate is processed by a CNC milling machine or CNC machining center; and the grinding wheel speed, table movement speed, depth of cut, and number of grinding passes when the chip contact bottom surface is ground by a CNC surface grinder using a precision grinding wheel.

[0006] Therefore, the optimization process for heat sinks designed to meet the thermal requirements of SoC chips typically involves analyzing the relationships between chip thermal characteristics, heat sink structural parameters, thermal contact conditions, and thermal performance evaluation results. Existing technologies already employ approaches that optimize heat sink structures and related thermal parameters by identifying key factors affecting chip junction temperature or thermal performance and combining these with experimental design, simulation analysis, surrogate models, or optimization algorithms.

[0007] For example, the Chinese invention patent application CN119783616A discloses a low-cost thermal management method for automotive SoC chips, which includes: S1: SoC chip analysis: Analyzing the SoC chip using a QFD analysis system to identify key factors affecting the SoC chip junction temperature, scoring each factor in the form of a House of Quality to obtain the importance ratio of each factor; S2: Key factor screening: Selecting the top four factors with the highest importance ratio based on the data reflected in the House of Quality, and labeling the selected key factors; S3: Key factor model design: Including DOE simulation model grouping design, removal of insignificant factors, response surface analysis, and response optimization. Thus, by designing and screening key factors through experimental analysis, conducting multiple sets of experiments on the key factors, and judging based on the experimental data and generated charts, the optimal solution that satisfies the various coefficients of the heat sink is obtained with the aim of reducing costs.

[0008] For example, Chinese invention patent application CN115659847B discloses a heat sink optimization method, apparatus, readable storage medium, and electronic device, which includes: establishing an optimal surrogate model for an electronic product with a heat sink; determining an objective function for optimization design based on the optimal surrogate model; and using an improved particle swarm optimization algorithm to optimize the optimal surrogate model according to the objective function to obtain the optimization result.

[0009] In the prior art, when configuring a heat sink for a SoC chip, the design and thermal performance evaluation are usually carried out based on the chip's overall power consumption, target junction temperature, installation space, and preset structural parameters of the heat sink, as well as the size of the heat sink substrate, the structure of the heat sink fins, the configuration of the thermal interface material, and the assembly clamping conditions.

[0010] However, during the optimization and adjustment of SoC heat sink production parameters based on thermal performance simulation results, the actual heat transfer path and interface heat transfer conditions may differ due to changes in the functional tasks, data processing intensity, and continuous load level of the SoC chip under the target application scenario. This is also affected by the connection and forming state between the heat sink substrate and the heat sink fins in the current production batch, as well as the spreading state of the thermal interface material in the contact area. Consequently, the thermal performance simulation results may not accurately reflect the actual heat dissipation capacity of the current batch of heat sinks. This makes it difficult to accurately determine the corresponding process parameters that cause the heat dissipation performance deviation and their adjustment direction when optimizing the heat sink production process parameters.

[0011] The specific reason is that existing thermal performance simulation analyses typically use the heatsink's preset structural dimensions, standard connection states, ideal interface bonding states, and fixed assembly clamping conditions as the modeling basis. Based on the model's output of temperature rise distribution, contact thermal resistance, and heat dissipation capacity evaluation results, relevant parameters during the heatsink production process are selected or modified. However, for actual manufactured SoC heatsinks, the connection and forming state between the heatsink substrate and the heatsink fins affects the effective path of heat transfer from the substrate to the fins, and the spread range and thickness uniformity of the thermally conductive interface material affect the interface heat transfer capacity between the chip package surface and the heatsink substrate. Existing methods struggle to uniformly convert the connection state data and interface material configuration data generated during the production process into model input parameters capable of correcting the heat transfer path, interface contact thermal resistance, and heat transfer boundary conditions in the simulation model. Furthermore, it is difficult to correlate the thermal input data corresponding to the target operating state with the actual heat transfer boundary data formed in the current batch of heatsinks, leading to the simulation model still evaluating different production parameter schemes according to preset states.

[0012] Therefore, when optimizing the heat sink connection forming parameters and thermal interface material configuration parameters for the target SoC chip, the actual heat diffusion path state and interface contact heat transfer state formed by the heat sink in the current batch may not be accurately mapped to the simulation model. This makes it difficult for the production parameter scheme determined based on the simulation evaluation results to improve the contact thermal resistance, heat diffusion uniformity and temperature rise control effect in the actual heat dissipation process, thereby reducing the accuracy and reliability of the SoC heat sink production process parameter optimization results. Summary of the Invention

[0013] This invention provides a data-driven adaptive optimization method for heat sink manufacturing process parameters. The technical solution is as follows: The method involves acquiring thermal input data of the target SoC chip under target operating conditions, and acquiring manufacturing process parameter data and production status detection data for the SoC heat sink in the current production batch. Based on the heat transfer relationship between the target SoC chip, heat sink substrate, and heat sink fins, the thermal input data and production status detection data are correlated to generate thermal transfer status data. The loading data of the initial thermal performance simulation model based on the SoC heat sink is corrected according to the thermal transfer status data, and a target thermal performance simulation model corresponding to the actual thermal transfer status of the SoC heat sink in the current production batch is obtained based on the acquired boundary parameters. Based on the target thermal performance simulation model, the thermal diffusion uniformity corresponding to different candidate manufacturing process parameter schemes is simulated and evaluated. The target manufacturing process parameter scheme is determined from the candidate manufacturing process parameter schemes based on the simulation evaluation results. The target manufacturing process parameter scheme is used for the production processing of subsequent production batches of SoC heat sinks, and the corresponding thermal performance detection results are obtained. The boundary parameters and / or the candidate manufacturing process parameter schemes corresponding to subsequent production batches are updated based on the thermal performance detection results.

[0014] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0015] 1. The data-driven adaptive optimization method for heat sink manufacturing process parameters provided by this invention acquires the thermal input data of the target SoC chip under the target operating conditions, and acquires the manufacturing process parameter data and production status detection data of the SoC heat sink in the current production batch. Based on the heat transfer relationship between the target SoC chip, the heat sink substrate, and the heat sink fins, the thermal input data and production status detection data are correlated to generate thermal transfer status data. This helps to unify the regionalized thermal input state formed by the chip's operating load, the interface spreading state formed after the actual production of the heat sink, and the connection state between the substrate and the fins into the same thermal transfer path for characterization, thereby improving the accuracy of identifying the actual thermal transfer state of the SoC heat sink in the current production batch. The loading data of the initial thermal performance simulation model based on the SoC heat sink is corrected based on the thermal transfer status data, and the correlation between the actual thermal transfer state of the SoC heat sink in the current production batch is obtained based on the acquired boundary parameters. The corresponding target thermal performance simulation model helps to transform the simulation model from an ideal structural state to an actual state model that reflects the interface contact thermal resistance and connection heat transfer state of the current batch, thereby improving the reliability of the temperature rise distribution and heat diffusion uniformity prediction results. Based on the target thermal performance simulation model, the heat diffusion uniformity corresponding to different candidate production process parameter schemes is simulated and evaluated. The target production process parameter scheme is determined from the candidate production process parameter schemes according to the simulation evaluation results, which is beneficial to pre-screen process parameter combinations that can improve heat dissipation performance before actual production. The target production process parameter scheme is used for the production processing of subsequent batches of SoC heat sinks, and the corresponding thermal performance test results are obtained. The boundary parameters and / or the candidate production process parameter schemes corresponding to subsequent production batches are updated according to the thermal performance test results, which is beneficial to form a closed-loop feedback between simulation optimization, production execution, physical inspection and model correction, and improve the adaptability of production parameters and the consistency of heat dissipation performance in subsequent batches.

[0016] 2. This invention combines the connection forming state data and thermal interface material spreading state data corresponding to the same sample identifier to form production state detection data. This production state detection data is then linked and stored with the initial production process parameters corresponding to the same sample identifier. Compared to traditional methods, its advantages lie in the fact that traditional methods typically only record processing parameters or only perform finished product thermal performance testing, making it difficult to determine whether abnormal temperature rise is caused by insufficient thermal interface material coverage, insufficient continuity between the substrate and fins, or other process factors. This invention binds the initial production process parameters to the connection forming state and thermal interface material spreading state through sample identifiers. This allows for tracing back to adjustable parameters such as coating speed, welding power, or welding scanning speed when interface heat transfer deviations or connection heat transfer deviations are subsequently detected, thereby improving the targeting and feasibility of adjusting production process parameters.

[0017] 3. By loading the set interface heat transfer boundary parameters, set connection heat transfer boundary parameters, and corresponding thermal input data into the initial thermal performance simulation model, a target thermal performance simulation model corresponding to the actual heat transfer state of the current production batch of SoC heat sinks is obtained. The simulation results of the temperature rise distribution and heat diffusion uniformity of the current production batch of SoC heat sinks output by the target thermal performance simulation model are read, and at least one of the interface heat transfer deviation path and the connection heat transfer deviation path is further determined as the target heat transfer path. Compared with traditional methods, its advantage lies in the fact that traditional thermal performance evaluation often only obtains the overall temperature rise or overall thermal resistance results, making it difficult to distinguish the source of anomalies. This invention evaluates the temperature rise of the bottom heated area and the fin connection associated area separately through the target thermal performance simulation model, which can identify the interface heat transfer deviation between the target SoC chip package surface and the heat dissipation substrate, as well as the connection heat transfer deviation between the heat dissipation substrate and the heat dissipation fins. This allows subsequent process parameter adjustments to correspond to the thermal interface material configuration process and the substrate fin connection forming process respectively, avoiding unnecessary adjustments to irrelevant processes.

[0018] 4. By reading the simulated temperature rise and simulated thermal diffusion uniformity results corresponding to the target production process parameter scheme in the target thermal performance simulation model, and comparing the actual temperature rise and actual thermal diffusion uniformity test results with the corresponding simulation results, the thermal performance deviation results are obtained. When the thermal performance deviation results meet the preset allowable deviation conditions, the target production process parameter scheme is determined as the confirmed production process parameter scheme for subsequent production batches. Compared with existing technologies, its advantage lies in the ability to verify the simulation optimization results using physical test results, avoiding the direct determination of production parameters based solely on a single simulation result. By comparing the deviations between the actual temperature rise and the simulated temperature rise, and between the actual thermal diffusion uniformity and the simulated thermal diffusion uniformity, it can be determined whether the boundary correction parameters in the target thermal performance simulation model match the actual production state.

[0019] 5. When the target SoC chip has multiple thermal contact areas with different thermal input intensities under the target operating conditions, the maximum thermal performance deviation result is obtained. When the maximum thermal performance deviation result meets the preset maximum allowable deviation condition, the target production process parameter scheme is determined as the confirmed production process parameter scheme for subsequent production batches. Compared with the existing technology, its advantage lies in highlighting the impact of high heat input areas and major uneven heat diffusion areas on heat dissipation reliability. By comparing the actual maximum temperature rise value with the simulated maximum temperature rise value, and the difference between the actual maximum temperature rise value and the difference between the simulated maximum temperature rise value, the evaluation focus can be concentrated on local hot spots and areas with the most uneven heat diffusion, thereby improving the reliability of screening production process parameter schemes under complex operating conditions. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart of a data-driven adaptive optimization method for heat sink manufacturing process parameters provided in an embodiment of this application;

[0022] Figure 2 This is a schematic diagram illustrating the generation of production status detection data provided in an embodiment of this application;

[0023] Figure 3 This is a schematic diagram illustrating the generation of the target thermal performance simulation model provided in the embodiments of this application;

[0024] Figure 4 This is a schematic diagram of the generation of the target heat transfer path provided in an embodiment of this application. Detailed Implementation

[0025] Embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of the present disclosure are shown in the drawings, it should be understood that embodiments of the present disclosure may be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the present disclosure.

[0026] It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure. In the description of the embodiments of this disclosure, the term "comprising" and similar terms should be understood as open-ended inclusion, i.e., "including but not limited to". The term "based on" should be understood as "at least partially based on". The term "one embodiment" or "this embodiment" should be understood as "at least one embodiment". The terms "first", "second", etc., may refer to different or the same objects.

[0027] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0028] Example 1, as Figure 1The diagram shows a flowchart of a data-driven adaptive optimization method for heat sink manufacturing process parameters provided in this application embodiment. The method includes the following steps: S1, acquiring the thermal input data of the target SoC chip under the target operating condition, and acquiring the manufacturing process parameter data and production status detection data corresponding to the SoC heat sink in the current production batch. Based on the heat transfer relationship between the target SoC chip, the heat sink substrate, and the heat sink fins, the thermal input data is correlated with the production status detection data to generate thermal transfer status data characterizing the actual thermal transfer status of the SoC heat sink in the current production batch; S2, based on the thermal transfer status data, the initial... S3. The loading data of the thermal performance simulation model is corrected, and a target thermal performance simulation model corresponding to the actual heat transfer state of the current production batch of SoC heat sink is obtained based on the acquired boundary parameters; S4. Based on the target thermal performance simulation model, the heat diffusion uniformity corresponding to different candidate production process parameter schemes is simulated and evaluated, and the target production process parameter scheme is determined from the candidate production process parameter schemes according to the simulation evaluation results; S5. The target production process parameter scheme is used for the production processing of subsequent production batches of SoC heat sinks, and the corresponding thermal performance test results are obtained. The boundary parameters and / or the candidate production process parameter schemes corresponding to subsequent production batches are updated according to the thermal performance test results.

[0029] In one specific embodiment, the target SoC chip is installed in an edge computing terminal and performs image acquisition, target recognition inference, and communication upload tasks under the target operating conditions, with a task duration of 600 seconds. Based on the correspondence between the target SoC chip package surface and the contact bottom surface of the SoC heat sink, the contact bottom surface of the heat sink is divided into four bottom surface heating regions A1 to A4. The heat input corresponding to each region is determined by the heat sink thermal matching model. Among them, region A1 has the highest heat flux density, which is 20,000 watts per square meter, with an operating time of 600 seconds, and is used to characterize this region as the main heat dissipation area under the target operating conditions.

[0030] The sample SoC heatsink H001 from the current production batch was obtained. Its manufacturing process parameters include a laser welding power of 420 watts, a welding scanning speed of 18 mm / s, a thermal grease application amount of 0.32 g, and an application speed of 25 mm / s. Industrial vision inspection equipment was used to inspect the connection area between the heatsink substrate and the heatsink fins. The effective continuous connection length of the fin connection associated region F1, corresponding to the heat diffusion path in region A1, was found to be 15.6 mm, which is less than the preset continuous connection length threshold of 18 mm. Therefore, region F1 was determined to be in a state of insufficient connection continuity. The thermal grease spreading state was inspected using a 3D contour measurement device. The thermal grease coverage ratio in region A1 was found to be 80%, indicating a heat transfer gap. Therefore, the heat flux density corresponding to region A1, the thermal grease coverage state, and the connection continuity state corresponding to region F1 were correlated to generate the heat transfer state data corresponding to the current production batch.

[0031] An initial thermal performance simulation model was established based on the preset structural parameters of the SoC heatsink, wherein the initial interface contact thermal resistance corresponding to each bottom heated area was set to 1.5 × 10⁻⁶. 4 The initial equivalent connection thermal resistance per square meter (Kelvin per watt) for each fin-connected region is set to 0.8 × 10⁻⁶. 4 Kelvin per watt per square meter. Due to the presence of heat transfer deficiencies in region A1, the interfacial contact thermal resistance of region A1 is adjusted to 2.4 × 10⁻⁶ based on the thermal grease coverage. 4 Kelvin per watt per square meter; Due to insufficient continuity in region F1, the equivalent connection thermal resistance of region F1 is corrected to 1.5 × 10⁻⁶ watts based on its effective continuous connection length. 4 Kelvin per square meter per watt. The corrected boundary parameters and the heat input data corresponding to each heated area on the bottom surface are loaded into the initial thermal performance simulation model to obtain the target thermal performance simulation model.

[0032] Simulations using the target thermal performance model yielded a temperature rise of 46.8 degrees Celsius for region A1 and 32.4 degrees Celsius for region F1. Since the temperature rise in region A1 exceeds the preset allowable temperature rise threshold of 40 degrees Celsius, and the temperature rise in region F1 exceeds the preset allowable temperature rise threshold for connection association of 30 degrees Celsius, it is determined that there is a heat dissipation performance deviation in the heat transfer path corresponding to A1 to F1.

[0033] The simulation results and current production process parameters were input into the heat sink production process parameter readjustment model to generate multiple candidate production process parameter schemes. Candidate Scheme 1 only increased the amount of thermal grease applied; Candidate Scheme 2 only increased the laser welding power; and Candidate Scheme 3 simultaneously adjusted the amount of thermal grease applied from 0.32 grams to 0.38 grams, the application speed from 25 mm / s to 20 mm / s, the laser welding power from 420 watts to 450 watts, and the welding scanning speed from 18 mm / s to 16 mm / s. Simulation evaluation showed that the temperature rise in region A1 corresponding to Candidate Scheme 3 was 37.6 degrees Celsius, and the temperature rise in region F1 was 27.9 degrees Celsius, both meeting the preset heat dissipation evaluation conditions. Therefore, Candidate Scheme 3 was selected as the target production process parameter scheme.

[0034] The target manufacturing process parameters were applied to the production of the sample SoC heatsink H101 in subsequent production batches, and thermal performance was tested under the same heat input conditions. The actual temperature rise in region A1 was 40.5 degrees Celsius, and the actual temperature rise in region F1 was 29.1 degrees Celsius. Comparing the actual test results with the corresponding simulation results, the temperature rise deviation in region A1 was found to be 2.9 degrees Celsius, which is greater than the preset allowable deviation threshold of 2 degrees Celsius. Therefore, based on the interface spreading state and connection continuity state obtained from subsequent tests, the interface contact thermal resistance correction value and the equivalent connection thermal resistance correction value were updated, and the target manufacturing process parameters were regenerated.

[0035] The updated target manufacturing process parameters include a thermal grease application amount of 0.41 grams, an application speed of 18 mm / s, a laser welding power of 455 watts, and a welding scanning speed of 15.5 mm / s. After applying the updated parameters to the production of the next batch of sample SoC heatsinks H201, the deviation between the actual and simulated temperature rise values ​​in region A1 was found to be 0.7 degrees Celsius, and the corresponding deviation in region F1 was 0.5 degrees Celsius. Both meet the preset allowable deviation conditions. Therefore, the updated target manufacturing process parameters are determined as the confirmed manufacturing process parameters for subsequent production batches.

[0036] The explanation will be based on steps S1-S4 provided in this application. First, based on step S1:

[0037] The specific process for acquiring production status detection data is as follows: acquire the target running task, data processing load, and task duration corresponding to the target SoC chip, and combine them to define the running condition data under the target running condition; under the target running condition, based on the pre-built heat sink thermal matching model, perform the input heat sink thermal matching model operation on the running condition data to obtain the heat input intensity, heat input duration, and heat input distribution location corresponding to each heat contact area, and use them as heat input data.

[0038] Obtain the batch identifier and sample identifier corresponding to each sample SoC heatsink in the current production batch, and read the initial production process parameters of the corresponding sample SoC heatsink from the historical production equipment control record based on the batch identifier and sample identifier. The initial production process parameters include at least one of the connection temperature, heat preservation duration, welding power and welding scanning speed corresponding to the connection and forming of the heat sink substrate and heat sink fins by the brazing furnace or laser welding equipment, and at least one of the coating amount, coating speed and bonding pressure corresponding to the thermal interface material configuration by the thermal grease automatic dispensing machine or thermal pad bonding equipment.

[0039] like Figure 2 The diagram illustrates the generation of production status detection data provided in this application. Based on industrial vision inspection equipment, such as a high-resolution area array industrial camera and an image processing terminal connected to the area array industrial camera, the connection status of the heat sink substrate and heat sink fin connection area of ​​each sample SoC heat sink in the current production batch is detected. This yields connection forming status data characterizing the continuity of the connection area. This quantifies the continuity of the connection between the substrate and fins into calculable data, enabling the connection heat transfer capability when the heat sink substrate transfers heat to the heat sink fins to be identified and utilized by subsequent thermal performance simulation models. This avoids judging the heat sink connection quality solely based on appearance or structural dimensions. The connection forming status data is generated by acquiring the connection appearance image of the exposed connection area between the heat sink substrate and heat sink fins using an area array industrial camera. The connection boundary is extracted from the connection appearance image based on the preset connection area position, and the length of the region continuously forming an effective connection along the predetermined connection direction is identified. The image processing terminal may include an industrial computer, an embedded vision processor, an edge computing terminal, etc. Based on industrial vision inspection equipment and three... A 3D contour measurement device detects the spread of thermal interface material on the contact bottom surface of a heat sink, obtaining thermal interface material spread data to characterize the coverage area of ​​the thermal interface material when thermal grease, thermal gel, or thermal pads have been applied or attached and the heat sink has not yet been pressed and assembled with the target SoC chip package surface. This data can identify whether there is insufficient coverage or localized missing parts of the thermal interface material in the chip contact area, allowing for a quantitative representation of the interface heat transfer state between the chip package surface and the heat sink substrate contact bottom surface. This improves the ability to identify the risk of increased local contact thermal resistance. The thermal interface material spread data can be specifically represented by the area actually covered by the thermal interface material in each thermal contact area on the heat sink contact bottom surface. The 3D contour measurement device includes a structured light 3D contour scanner or a laser confocal 3D contour measuring instrument. The connection forming state data and thermal interface material spread state data corresponding to the same sample identifier are combined to form production state detection data, which is then associated and stored with the initial production process parameters corresponding to the same sample identifier.

[0040] By acquiring the thermal input data of the target SoC chip under the target operating conditions, and combining it with the initial production process parameters, connection and forming status data, and thermal interface material spreading status data of the current production batch of SoC heat sinks, subsequent analysis can no longer rely solely on the heat sink's design structure parameters, but can simultaneously consider the thermal input state formed by the actual operating load of the target chip and the thermal conductivity structure state formed after the actual production of the current batch of heat sinks, thereby improving the authenticity and relevance of the thermal transfer state characterization.

[0041] By reading the initial manufacturing process parameters of the corresponding sample SoC heatsink based on batch and sample identifiers, and associating and storing these initial manufacturing process parameters with production status detection data, a correspondence between production equipment control parameters and actual forming states can be established. For example, laser welding power and welding scanning speed can be correlated with the effective continuous connection length between the heatsink substrate and heatsink fins, and the amount of thermal grease applied, application speed, and attachment pressure can be correlated with the actual coverage area of ​​the thermal interface material. This provides a data basis for subsequently determining the source of process parameters corresponding to abnormal heat dissipation performance.

[0042] Here, "target running task" refers to the functional processing tasks that the target SoC chip needs to perform during the operation of a predetermined application device, such as device control instruction processing tasks and communication data processing tasks; "data processing load" refers to the amount of data processing work undertaken by the target SoC chip during the execution of the corresponding target running task, which can be determined by at least one of processor core utilization, memory access frequency, interface data transfer rate, and / or operating power consumption; "task continuous running time" refers to the length of time the target SoC chip continuously executes the target running task according to the corresponding data processing load; "target SoC chip" refers to the system-on-chip that needs to be thermally adapted to the current batch of SoC heat sinks, serving as the target heat source object for generating thermal input data; and "target operating condition" refers to the working state of the target SoC chip to be adapted to perform a set combination of functional tasks in a predetermined application device and continuously operate according to the corresponding data processing load level. The device refers to an electronic device equipped with a target SoC chip and requiring the transfer and release of heat generated during the operation of the target SoC chip to the external environment through an SoC heat sink, including at least one of industrial control terminals, communication processing terminals, edge computing terminals, and smart electronic terminals; the thermal contact area refers to multiple thermally conductive contact area units obtained by dividing the bottom surface of the heat sink according to the projection correspondence between the surface of the target SoC chip package and the bottom surface of the SoC heat sink in the assembled state; the heat input intensity refers to the heat flow transfer rate per unit area of ​​the target SoC chip inputting heat to the bottom surface of the heat sink in the corresponding thermal contact area under the target operating conditions, specifically expressed as the heat flow density of the corresponding thermal contact area; the heat input duration refers to the continuous duration for which the corresponding thermal contact area maintains the heat input level; the heat input distribution location refers to the area number and location range of the corresponding thermal contact area within the coordinate range of the bottom surface of the heat sink.

[0043] Based on the heat transfer relationship between the target SoC chip, heat sink substrate, and heat sink fins, the thermal input data is correlated with the production status detection data. The specific process is as follows: Based on the preset assembly positioning relationship between the target SoC chip package surface and the SoC heat sink contact bottom surface, the bottom surface heated area corresponding to each thermal contact area on the heat sink contact bottom surface is obtained; according to the thermal input distribution position corresponding to each thermal contact area, the corresponding thermal input intensity and thermal input duration are bound to the bottom surface heated area to obtain the regional thermal input state corresponding to each bottom surface heated area. The specific process is as follows: Read the region number, planar coordinate range, and region boundary range corresponding to any thermal contact area, and record the thermal input intensity and thermal input duration corresponding to the thermal contact area to the corresponding bottom surface heated area. The thermal regions provide each bottom-side heated area with corresponding thermal input state data, including the heat flux density and duration of heat flux application. By using the preset assembly positioning relationship between the target SoC chip package surface and the SoC heat sink contact bottom surface, each thermal contact area is mapped to a bottom-side heated area on the heat sink contact bottom surface. The thermal input intensity and duration are bound to the corresponding bottom-side heated area, which clarifies the actual position of different thermal input sources on the heat sink, thus establishing an accurate correspondence between thermal input data and heat sink structural areas. The thermal input distribution location includes the area number, coordinate position, and boundary coverage of the thermal contact area when it acts on the heat sink contact bottom surface, used to determine the starting area for heat to enter the heat sink contact bottom surface.

[0044] Based on the area actually covered by the thermally conductive interface material in each thermal contact area from the thermal interface material spread state data, the effective coverage range of the interface corresponding to each bottom heated area is determined. This effective coverage range is then correlated with the corresponding regional thermal input state to obtain the interface heat transfer state from the target SoC chip package surface to the bottom contact surface of the heat dissipation substrate. Specifically, when the area actually covered by the thermally conductive interface material in a bottom heated area is smaller than the corresponding bottom heated area area, the uncovered location is defined as a heat transfer gap in the corresponding interface heat transfer state. When the area actually covered by the thermally conductive interface material in a bottom heated area is greater than or equal to the corresponding bottom heated area area, the bottom heated area is defined as a continuous interface coverage area. The effective coverage range is correlated with the corresponding regional thermal input state. For the same bottom heated area, the heat flux density and... The duration of heat flux density is linked to the effective coverage of the interface to characterize whether the heat input to the bottom heated area can form a continuous interface transfer range through the thermally conductive interface material. The substrate outline, heat dissipation fin arrangement, and connection area between the heat dissipation substrate and the heat dissipation fins are obtained. Taking each bottom heated area as the heat input starting area, the substrate diffusion association area and the corresponding fin connection association area are determined according to the heat diffusion direction extending from the contact bottom to the heat dissipation fin connection area inside the heat dissipation substrate. By associating the thermally conductive interface material spread state with the regional heat input state, it is possible to determine whether the high heat input area has a continuous thermally conductive interface material coverage range, and identify the heat transfer missing location when the coverage is insufficient. This is helpful in judging whether the local temperature rise anomaly is caused by insufficient thermally conductive interface material configuration and improves the accuracy of interface heat transfer anomaly location.

[0045] Based on the connection formation state data, the connection continuity state corresponding to each fin connection association area is determined. The specific process is as follows: The effective continuous connection length of the fin connection association area corresponding to the connection formation state data is read. The effective continuous connection length is compared with the preset continuous connection length threshold of the corresponding fin connection association area to obtain the connection continuity state used to characterize the actual continuous connection degree of the corresponding connection area. The effective continuous connection length represents the actual length, monitored by a high-resolution area array industrial camera and image processing terminal, in the fin connection association area where the heat dissipation substrate and heat dissipation fins continuously form an effective connection state along a predetermined connection direction. When the effective continuous connection length of the fin connection association area... When the continuous connection length is greater than or equal to the preset continuous connection length threshold, the corresponding fin connection associated area is determined to be in a state of continuous connection satisfaction; otherwise, the corresponding fin connection associated area is determined to be in a state of insufficient continuous connection, and the corresponding connection interruption position and connection interruption length are recorded. By obtaining the heat dissipation substrate outline, heat dissipation fin arrangement position, and connection area position between the heat dissipation substrate and heat dissipation fins, and determining the substrate diffusion associated area and fin connection associated area corresponding to each bottom heated area, a heat transfer path can be established from the chip heat input area to the heat dissipation substrate diffusion area, and then to the fin connection area, thus expanding the heat dissipation performance analysis from single area judgment to complete heat transfer path judgment.

[0046] The preset assembly positioning relationship represents the spatial correspondence between the target SoC chip package surface or corresponding thermal cover and the contact bottom surface of the SoC heatsink in a predetermined assembly state. This spatial correspondence is predetermined based on the target SoC chip package dimensions, heatsink contact bottom surface dimensions, assembly positioning reference points, mounting holes, positioning boundaries, and assembly direction. It is used to map the thermal contact area on the target SoC chip package surface to the corresponding position area on the heatsink contact bottom surface. The substrate outline represents the outer boundary position, contact bottom surface position, fin connection surface position, and substrate thickness range of the heatsink substrate in a preset structural coordinate system. The heatsink fin arrangement position represents the mounting area number, arrangement direction, and position range of each heatsink fin or fin group on the heatsink substrate connection surface. The connection area position between the heatsink substrate and the heatsink fins represents the area number, coordinate range, and connection extension direction of the area where the bottom of each heatsink fin connects to the heatsink substrate connection surface.

[0047] The preset continuous connection length threshold represents the minimum effective connection length pre-set to meet the predetermined heat transfer continuity requirements between the heat dissipation substrate and the heat dissipation fins. It can be determined based on the designed connection extension length and allowable connection interruption range of the corresponding fin connection association area. For example, for a fin connection association area with a designed connection extension length of 20 mm, after testing a sample heat sink with qualified connection status and thermal resistance, temperature rise, and heat diffusion uniformity that meet the predetermined heat dissipation requirements, it is determined that the cumulative connection interruption length allowed in this connection area does not exceed 2 mm. Therefore, 18 mm can be determined as the preset continuous connection length threshold for the corresponding fin connection association area. When the detected effective continuous connection length is greater than or equal to 18 mm, it indicates that the connection area meets the predetermined heat transfer continuity requirements. When the detected effective continuous connection length is less than 18 mm, it indicates that there is insufficient connection continuity in this connection area, which needs to be treated as an abnormal factor in the connection heat transfer status during subsequent simulation analysis and production parameter optimization.

[0048] In one embodiment, the preset continuous connection length threshold can also be obtained by verifying the thermal performance of multiple sample SoC heat sinks with different connection states. Specifically, the effective continuous connection length and corresponding contact thermal resistance of each sample SoC heat sink are collected. The minimum effective continuous connection length of the sample whose thermal performance test results meet the predetermined qualification conditions is determined as the preset continuous connection length threshold. Alternatively, the minimum effective continuous connection length can be increased by a predetermined safety margin and then determined as the preset continuous connection length threshold. The thermal performance test results represent the contact thermal resistance detection value determined based on the correspondence between the temperature change of the simulated SoC heat sink and the set thermal power after the sample SoC heat sink is bonded to the simulated SoC chip heat source through a thermally conductive interface material and a preset thermal input condition is met. The predetermined qualification condition indicates that the contact thermal resistance detection value is less than or equal to the preset allowable thermal resistance upper limit.

[0049] It should be emphasized that other thresholds or preset values ​​provided in this application can be constructed using a similar method to the preset continuous connection length threshold, i.e., based on the design allowable range of the corresponding parameters, the test results of qualified samples, and the thermal performance verification results, which are predetermined.

[0050] By binding the continuous connection state to the substrate diffusion association region corresponding to the bottom heated area, the connection heat transfer state corresponding to the heat transfer from the heat dissipation substrate to the heat dissipation fins is obtained. This allows the connection heat transfer state when the heat dissipation substrate transfers heat to the heat dissipation fins to be included in the corresponding heat transfer path, thereby distinguishing between interface heat transfer anomalies and connection heat transfer anomalies in subsequent analysis and avoiding the confusion of heat dissipation performance deviations caused by different processes. For each bottom heated area, its corresponding regional heat input state, interface heat transfer state, substrate diffusion association region, and connection heat transfer state are combined to generate corresponding heat transfer path state data. The heat transfer path state data corresponding to multiple bottom heated areas are collected to generate heat transfer state data to characterize the actual heat transfer state of the SoC heat sink in the current production batch.

[0051] The aforementioned binding refers to data association processing that establishes a corresponding relationship between different types of data belonging to the same heat transfer path based on the area number, coordinate position, boundary coverage, and heat transfer path identifier of each bottom heated area extending from the bottom heated area to the fin connection association area. Specifically, according to the heat input distribution position corresponding to each heat contact area, the corresponding heat input intensity and heat input duration are bound to the bottom heated area, indicating that the heat flux density and its duration of action of the corresponding heat contact area are recorded as the heat input state of the bottom heated area. The heat flux density and its duration of action of the bottom heated area are bound to the effective coverage area of ​​the interface, indicating that the actual coverage area of ​​the thermally conductive interface material and the location of heat transfer deficiency in the same bottom heated area are recorded in correspondence with the heat flux density and duration of action of the area. The continuous connection state is bound to the substrate diffusion association area corresponding to the bottom heated area, indicating that the effective continuous connection length, connection interruption position, or connection interruption length after reaching the fin connection association area along the corresponding heat transfer path are recorded as the connection heat transfer conditions when heat is transferred from the heat dissipation substrate to the heat dissipation fins in the heat transfer path.

[0052] In this embodiment, the regionalized thermal input data of the target SoC chip under target operating conditions, the manufacturing process parameter data corresponding to the current production batch of the SoC heatsink, the thermal interface material spreading status data, and the connection status data between the heatsink substrate and the heatsink fins are correlated and processed to construct a heat transfer path status data from the chip's thermal input area to the heatsink contact bottom surface, and then diffused through the heatsink substrate to the heatsink fin connection area. This processing method allows the heat input source, interface heat transfer conditions, and connection heat transfer conditions to be expressed correspondingly in the same heat transfer path. This is beneficial for accurately identifying the difference between the actual heat transfer state and the ideal heat transfer state of the current batch of SoC heatsinks, and provides reliable data support for subsequent boundary correction and manufacturing process parameter optimization based on thermal performance simulation models. This improves the pertinence, accuracy, and batch adaptability of SoC heatsink manufacturing process parameter optimization.

[0053] Step S2:

[0054] like Figure 3 The diagram shows the generation of the target thermal performance simulation model provided in this application. The specific process for obtaining the target thermal performance simulation model corresponding to the actual heat transfer state of the current production batch of SoC heat sinks is as follows: Heat transfer state data is read, and based on the continuous interface coverage area or heat transfer gap location within each bottom heated area, corresponding interface heat transfer boundary correction parameters are generated. This transforms the influence of the actual spread state of the thermally conductive interface material on the heat transfer capability between the chip package surface and the heat dissipation substrate into interface contact thermal resistance parameters recognizable by the initial thermal performance simulation model. This prevents the simulation model from solely calculating based on the ideal coverage state, thereby improving the accuracy of the interface heat transfer state characterization. Specifically, when a continuous interface coverage area is detected, the correction value of the interface heat transfer boundary parameters is not obtained; the initial interface heat transfer boundary parameters of the corresponding interface area in the initial thermal performance simulation model are maintained. As a set interface heat transfer boundary parameter, it can avoid unnecessary corrections to areas that already meet the thermal interface material coverage requirements, reduce the computational load of the model, and maintain the stability of the simulation boundary conditions corresponding to the normal heat transfer area. When a heat transfer deficiency location is detected, the thermal interface material spread state data corresponding to that location is substituted into the pre-constructed interface heat transfer boundary parameter correction model in the next preset time period to obtain the correction value of the interface heat transfer boundary parameter. This value is then used as the set interface heat transfer boundary parameter. Based on the location, area, or degree of insufficient coverage of the heat transfer deficiency area, the interface contact thermal resistance of the corresponding area can be specifically corrected, so that the simulation model can reflect the problem of increased contact thermal resistance caused by local deficiencies in the thermal interface material. The interface heat transfer boundary parameter includes the interface contact thermal resistance between the target SoC chip package surface or the corresponding thermal cover plate and the bottom surface of the heat dissipation substrate.

[0055] The system reads heat transfer state data and generates correction values ​​for corresponding connection heat transfer boundary parameters based on the connection continuity satisfaction or insufficient connection continuity states of each fin connection associated region. This transforms the actual connection continuity between the heat sink and the heat sink fins into connection heat transfer boundary conditions in the simulation model, allowing the influence of connection state on the heat transfer from the heat sink to the heat sink fins to be quantified. Specifically: when a connection continuity satisfaction state is detected, no correction value for the connection heat transfer boundary parameters is acquired; the initial connection heat transfer boundary parameters of the corresponding connection region in the initial thermal performance simulation model are maintained and used as the set connection heat transfer boundary parameters, avoiding over-correction of fin connection regions with normal connection states; when a connection continuity satisfaction state is detected, the system does not acquire correction values ​​for the connection heat transfer boundary parameters. When the connection is deemed insufficient, the connection interruption location and length corresponding to the fin connection associated area are input into the correction model of the connection heat transfer boundary parameters in the next preset time period. The correction value of the connection heat transfer boundary parameters output by the model is used as the set connection heat transfer boundary parameters. Based on the impact of connection interruption on the continuity of heat transfer path, the equivalent connection thermal resistance of the corresponding area can be corrected, thereby improving the model's ability to express the limited heat transfer state from the substrate to the fins. The boundary parameters and the corresponding thermal input data are loaded into the initial thermal performance simulation model to obtain the target thermal performance simulation model corresponding to the actual heat transfer state of the current production batch of SoC heat sinks. The boundary parameters include the set interface heat transfer boundary parameters and the set connection heat transfer boundary parameters.

[0056] In one specific embodiment, the initial thermal performance simulation model can be constructed based on the electronic packaging and heat sink thermal contact resistance model in existing finite element thermal simulation software, such as using the solid heat transfer model in the COMSOL Multiphysics heat transfer module. The solid heat transfer model is used to simulate and analyze the process of heat being transferred from the surface of the target SoC chip package to the heat dissipation substrate through the thermally conductive interface material, and further transferred to the heat dissipation fins and the external environment through the connection area between the heat dissipation substrate and the heat dissipation fins.

[0057] Specifically, the package outline dimensions, thermal contact area location on the package surface, and thermal cover location of the target SoC chip are obtained, and the three-dimensional structural data of the SoC heat sink to be simulated is obtained. The three-dimensional structural data includes the heat sink contact bottom surface dimensions, heat sink substrate outline dimensions and thickness, heat sink fin height and arrangement location, and connection area location between heat sink substrate and heat sink fins. The target SoC chip package surface, thermal interface material layer, heat dissipation substrate, and heat dissipation fins are imported into a solid heat transfer model according to a preset assembly position relationship to establish a three-dimensional heat transfer structure corresponding to the SoC chip and SoC heat sink. Based on the different heat transfer positions in the heat sink structure, the three-dimensional heat transfer structure is divided into regions. The area between the target SoC chip package surface and the bottom surface of the heat sink is divided into the interface heat transfer region, the area of ​​the heat dissipation substrate extending from the bottom surface of the heat sink to the heat dissipation fins is divided into the substrate diffusion region, the area where the heat dissipation substrate and the heat dissipation fins are connected is divided into the connection heat transfer region, and the surface of the heat dissipation fins in contact with the external environment is divided into the external heat dissipation region. At the same time, based on the mapping position of each thermal contact region on the bottom surface of the heat sink, the interface heat transfer region is divided into multiple simulation loading regions corresponding to the bottom surface heating region.

[0058] Input the thermal properties of the corresponding materials into the solid heat transfer model. Specifically, for the heat dissipation substrate and heat dissipation fins, input the thermal conductivity, density, and specific heat capacity of the aluminum alloy or copper alloy material used; for the thermal interface material layer, input the thermal conductivity and initial thickness of the thermal grease, thermal gel, or thermal pad; and for the connection area between the heat dissipation substrate and heat dissipation fins, configure the corresponding initial connection heat transfer parameters according to the preset connection continuity state.

[0059] Based on the preset interface bonding conditions and preset connection continuity conditions corresponding to the ideal production state of the SoC heat sink, an initial interface contact thermal resistance is set in the interface heat transfer area, and an initial equivalent connection thermal resistance is set in the connection heat transfer area. The initial interface contact thermal resistance is used to represent the degree of heat transfer resistance between the target SoC chip package surface and the contact bottom surface of the heat sink substrate when the thermally conductive interface material is continuously covered and the bonding state meets the preset requirements. The initial equivalent connection thermal resistance is used to represent the degree of heat transfer resistance during the process of heat transfer from the heat sink substrate to the heat sink fins when the connection continuity between the heat sink substrate and the heat sink fins meets the preset requirements.

[0060] Based on the heat input data corresponding to the target operating conditions, the heat flux density of each heated area on the bottom surface is set as the heat input boundary condition for that area in the model, and the continuous running time of the task is set as the simulation duration. According to the predetermined heat dissipation environment of the SoC heatsink, external convection heat transfer boundary conditions are set on the outer surface of the heatsink fins, and the preset ambient temperature or the steady-state temperature before heat input loading is set as the initial temperature. After completing the setting of geometry, material properties, region partitioning, initial contact thermal resistance, initial connection thermal resistance, heat input boundary, and external heat dissipation boundary, the model is meshed and transient heat transfer is solved to obtain the initial thermal performance simulation model based on the preset structural parameters of the SoC heatsink and the ideal heat transfer state.

[0061] In this embodiment, the thermal interface material spreading state, the continuous connection state between the heat dissipation substrate and the heat dissipation fins, and the regionalized thermal input data corresponding to the target SoC chip are uniformly converted into boundary parameters and loading data in the thermal performance simulation model. This achieves a mapping from production status detection data to thermal performance simulation boundary conditions. Through this mapping relationship, the initial thermal performance simulation model can be modified from a general model under ideal structural conditions to a target thermal performance simulation model that reflects the actual interface heat transfer conditions and connection heat transfer conditions of the current production batch. This improves the accuracy of the simulation results in predicting the actual temperature rise distribution, contact thermal resistance changes, and thermal diffusion uniformity, and provides a more reliable model basis for the simulation evaluation and optimization selection of subsequent candidate production process parameter schemes.

[0062] S3 Steps:

[0063] like Figure 4 The diagram shown is a schematic of the target heat transfer path generation provided in this application. By reading the simulation results of the temperature rise distribution and heat diffusion uniformity of the current production batch of SoC heat sinks output by the target thermal performance simulation model, the heat dissipation response of the current production batch of SoC heat sinks under the target operating conditions can be converted into comparable regional temperature rise data, so that the heat dissipation status of each bottom heated area and each fin connection area can be quantitatively evaluated. The simulation results include the temperature rise value corresponding to each bottom heated area and the temperature rise value corresponding to each fin connection area. The temperature rise value represents the difference between the simulation temperature of the corresponding area at the end of the preset simulation duration and the initial temperature of the area before the heat input is loaded after the boundary parameters and corresponding heat input data corresponding to the target operating conditions are loaded into the target thermal performance simulation model.

[0064] Based on simulation results, a target heat transfer path with heat dissipation performance deviation is determined. When the temperature rise value corresponding to any bottom heated area is greater than a preset allowable temperature rise threshold, the heat transfer path containing the bottom heated area and its corresponding interface heat transfer state is determined as an interface heat transfer deviation path; otherwise, the heat transfer path containing the bottom heated area and its corresponding interface heat transfer state is determined as an interface heat transfer normal path. When the temperature rise value corresponding to any fin connection associated area is greater than a preset connection associated allowable temperature rise threshold, the heat transfer path containing the fin connection associated area and its corresponding connection heat transfer state is determined as a connection heat transfer deviation path; otherwise, the heat transfer path containing the corresponding fin connection associated area and its corresponding connection heat transfer state is determined as a connection heat transfer normal path. At least one of the interface heat transfer deviation path and the connection heat transfer deviation path is determined as the target heat transfer path. When both the interface heat transfer normal path and the connection heat transfer normal path are detected, the corresponding initial production process parameters are determined as the target production process parameter scheme. This avoids unnecessary adjustments to production parameters that already meet heat dissipation requirements, reduces the number of repeated parameter adjustments and verifications of production equipment, and helps maintain the stability of the production process.

[0065] By comparing the temperature rise value corresponding to each heated area on the bottom surface with a preset allowable temperature rise threshold, it is possible to determine whether the interface heat transfer state between the target SoC chip package surface and the bottom surface of the heat dissipation substrate meets the heat dissipation requirements. When the temperature rise value of the heated area on the bottom surface exceeds the threshold, the corresponding heat transfer path is identified as the interface heat transfer deviation path, which is helpful in identifying local temperature rise anomalies caused by factors such as insufficient thermal interface material coverage and high interface contact thermal resistance.

[0066] By comparing the temperature rise value corresponding to the connection association area of ​​each fin with the preset allowable temperature rise threshold for connection association, it is possible to determine whether the connection heat transfer state meets the requirements when the heat sink substrate transfers heat to the heat sink fins. When the temperature rise value of the fin connection association area exceeds the threshold, the corresponding heat transfer path is identified as the connection heat transfer deviation path, which is helpful in identifying problems such as insufficient connection continuity between the heat sink substrate and the heat sink fins, large connection thermal resistance, or limited heat diffusion.

[0067] Based on the interface heat transfer state and / or connection heat transfer state corresponding to the target heat transfer path, the production process parameters to be adjusted are determined. The specific process is as follows: The simulation results corresponding to the target heat transfer path and the initial production process parameters are input into the heat sink production process parameter readjustment model based on machine learning to determine the production process parameters to be adjusted. The heat sink production process parameter readjustment model is obtained by training and validating a regression model, decision tree model, support vector regression model or backpropagation neural network model.

[0068] Furthermore, the heat sink thermal matching model, the interface heat transfer boundary parameter correction model, and the connection heat transfer boundary parameter correction model provided in this application are all pre-built based on machine learning models through separate training and verification. Specifically, the training sample data for the heat sink thermal matching model includes the operating condition data and thermal input data of sample SoC heat sinks that have completed thermal performance verification in historical production batches; the training sample data for the interface heat transfer boundary parameter correction model includes the thermal interface material spreading state data and the correction values ​​of the interface heat transfer boundary parameters of sample SoC heat sinks that have completed thermal performance verification in historical production batches; and the training sample data for the connection heat transfer boundary parameter correction model includes the connection interruption location, connection interruption length, and correction values ​​of the connection heat transfer boundary parameters of the fin connection association area of ​​sample SoC heat sinks that have completed thermal performance verification in historical production batches.

[0069] In one specific embodiment, the heat sink manufacturing process parameter readjustment model is constructed using a backpropagation neural network model. The model includes an input layer, a hidden layer, and an output layer. The input layer is used to receive the simulation results corresponding to the target heat transfer path, the production status detection results, and the current production process parameters. The first hidden layer is used to extract the correlation features between the degree of temperature rise anomaly and the interface coverage state or the continuous connection state. The second hidden layer is used to extract the correspondence between the correlation features and the adjustment direction of the production process parameters. The output layer is used to output the adjustment amount of the production process parameters to be adjusted.

[0070] Acquire sample SoC heat sinks from multiple historical production batches that have completed thermal performance verification, and record the simulation results, manufacturing process parameter data, and target manufacturing process parameters that meet the heat dissipation requirements after adjustment for each sample. Construct model training sample data, in which the simulation temperature rise value, interface coverage state and / or connection continuity state corresponding to the sample, as well as the manufacturing process parameters used before adjustment, are used as model input data, and the difference between the target manufacturing process parameters that meet the heat dissipation requirements after adjustment and the manufacturing process parameters before adjustment are used as output labels.

[0071] During training, the sample data is normalized and divided into training and validation sets. The weights and biases of the backpropagation neural network model are updated iteratively using the training set, gradually bringing the adjusted production process parameters output by the model closer to the corresponding output labels. The validation set is then used as input to the trained model to obtain the corresponding predicted parameter adjustments, and validation candidate production process parameter schemes are generated based on these adjustments. The interface heat transfer boundary parameters and / or connection heat transfer boundary parameters corresponding to the validation candidate production process parameter schemes are loaded into the thermal performance simulation model to obtain the validation simulation results. It is then determined whether the validation simulation results meet preset conditions. If the preset conditions are met, the trained model is designated as the discrete... If the heat exchanger production process parameters are readjusted and the model does not meet the preset conditions, training samples of the corresponding path type are added based on the target heat transfer path whose simulated temperature rise value in the verification samples still exceeds the corresponding allowable temperature rise threshold. The weights and biases of the backpropagation neural network model are iteratively updated until the verification simulation results meet the preset conditions. When the model iterative training times reach the preset maximum number of training times, if the verification simulation results still do not meet the preset conditions, a model training anomaly prompt is output. The preset conditions indicate that the simulated temperature rise value of each bottom surface heated area corresponding to the verification candidate production process parameter scheme is less than or equal to the preset allowable temperature rise threshold, and the simulated temperature rise value of each fin connection associated area is less than or equal to the preset connection associated allowable temperature rise threshold.

[0072] Based on the current parameter values ​​corresponding to the production process parameters to be adjusted, candidate production process parameter schemes are generated. Based on the candidate production process parameter schemes, the simulation results of temperature rise distribution and heat diffusion uniformity corresponding to the current production batch of SoC heat sinks output by the target thermal performance simulation model are reread. When the simulation results meet the preset diffusion uniformity conditions, the candidate production process parameter scheme is determined as the target production process parameter scheme; otherwise, a simulation evaluation alarm is sent. When the simulation results corresponding to the candidate production process parameter scheme meet the preset diffusion uniformity conditions, the candidate scheme is determined as the target production process parameter scheme, which can ensure that the selected parameter scheme can not only reduce local temperature rise, but also improve the uniformity of heat diffusion from the heat sink substrate to the heat sink fins. When the candidate scheme does not meet the requirements, a simulation evaluation alarm is sent, which can promptly indicate that the current candidate parameter adjustment is insufficient or the model evaluation is not up to standard, so as to facilitate the regeneration of candidate schemes or the verification of production equipment status. The preset diffusion uniformity conditions mean that the temperature rise value corresponding to each bottom heated area is less than or equal to the preset allowable temperature rise threshold, and the temperature rise value corresponding to each fin connection associated area is less than or equal to the preset connection associated allowable temperature rise threshold.

[0073] In this embodiment, the regional temperature rise results output by the target thermal performance simulation model are used to distinguish the thermal performance deviation of the SoC heat sink into interface heat transfer deviation and connection heat transfer deviation. Furthermore, different deviation paths are associated with the corresponding production process parameter adjustment objects, so that the thermal performance simulation results can directly serve the optimization of SoC heat sink production process parameters. This is beneficial to improving the screening efficiency of candidate production process parameter schemes, the accuracy of parameter adjustment, and the consistency of thermal performance of subsequent production batches.

[0074] Step S4:

[0075] The target production process parameters are distributed to the corresponding production equipment, such as an automatic thermal grease dispensing machine or a thermal pad attaching machine, so that subsequent batches of SoC heat sinks are produced according to the target production process parameters. Sample SoC heat sinks produced according to the target production process parameters are obtained and their thermal performance is tested to obtain the corresponding actual temperature rise and actual heat diffusion uniformity test results. The simulated temperature rise and simulated heat diffusion uniformity results corresponding to the target production process parameters in the target thermal performance simulation model are read, and the actual temperature rise and actual heat diffusion uniformity test results monitored by temperature sensors are compared with the corresponding simulation results to obtain the thermal performance deviation results, specifically including the difference between the actual temperature rise value and the simulated temperature rise value for each bottom heated area. The absolute value of the difference, and the absolute value of the difference between the actual temperature rise and the simulated temperature rise corresponding to the connected areas of each fin; determine whether the thermal performance deviation result meets the preset allowable deviation condition. If it does, the target production process parameter scheme is determined as the confirmed production process parameter scheme for the subsequent production batch. If it does not meet the condition, the correction values ​​of the corresponding interface heat transfer boundary parameters and / or connection heat transfer boundary parameters are updated to obtain the updated boundary parameters. Based on the updated boundary parameters, the target thermal performance simulation model is updated, and the target production process parameter scheme for the subsequent production batch is regenerated. If the number of updates to the target production process parameter scheme reaches the preset maximum number of updates, and the thermal performance deviation result still does not meet the preset allowable deviation condition, a radiator production parameter adjustment alarm is sent.

[0076] Among them, the actual temperature rise test result represents the actual temperature rise value corresponding to each bottom surface heated area, and the actual heat diffusion uniformity test result represents the actual temperature rise value corresponding to each fin connection associated area; the preset allowable deviation condition means that the absolute value of the difference between the actual temperature rise value and the simulated temperature rise value corresponding to each bottom surface heated area is less than or equal to the preset bottom surface temperature rise deviation threshold, and the absolute value of the difference between the actual temperature rise value and the simulated temperature rise value corresponding to the fin connection associated area is less than or equal to the preset connection area temperature rise deviation threshold.

[0077] In this embodiment, by distributing the target production process parameter scheme to the corresponding production equipment and conducting actual thermal performance testing on subsequent batches of SoC heat sinks, the parameter scheme obtained from the previous simulation optimization can be verified with the actual production results. By comparing the actual temperature rise test results, the actual heat diffusion uniformity test results, and the corresponding simulation results output by the target thermal performance simulation model, it can be determined whether the current boundary correction parameters accurately reflect the actual interface heat transfer state and connection heat transfer state. When the deviation between the two meets the preset allowable deviation conditions, the target production process parameter scheme is confirmed as an executable scheme for subsequent batches, which is beneficial to maintaining stable production parameters and consistent heat dissipation performance. When the deviation between the two does not meet the conditions, the interface heat transfer boundary parameter correction value and / or connection heat transfer boundary parameter correction value are updated in reverse according to the deviation results, and the target production process parameter scheme is regenerated based on the updated target thermal performance simulation model. This allows the SoC heat sink production parameters to no longer rely solely on a single simulation result or human experience for determination, but can be continuously corrected by combining the actual heat dissipation test results of subsequent batches, thereby improving the matching degree between the target thermal performance simulation model and the actual production state, and enhancing the accuracy, adaptability, and batch stability of SoC heat sink production process parameter optimization.

[0078] Taking a split aluminum fin heat sink used for surface heat dissipation of SoC chips as an example, the production process is introduced. First, aluminum alloy plates and aluminum alloy sheets are respectively fed to a CNC cutting machine and a precision shearing machine. The CNC cutting machine cuts the aluminum alloy sheet into a heat dissipation substrate blank according to the preset shape and size. The precision shearing machine cuts the aluminum alloy sheet into strip blanks for forming heat dissipation fins. Subsequently, the heat sink substrate blank is fed into a CNC machining center, where a face milling cutter rough-machines the outer contour, mounting holes, and bottom contact area corresponding to the SoC chip. The strip-shaped fin blank is then fed into a high-speed stamping machine and processed into a heat sink fin assembly with predetermined fin height, fin spacing, and arrangement pitch using a corrugated fin forming die or bending die. After processing the heat sink substrate and heat sink fins, a fin arrangement positioning fixture is used to arrange the heat sink fin assembly at a preset position on the connection surface of the heat sink substrate. A flux sprayer or solder paste applicator is then used to apply soldering material to the connection area between the heat sink substrate and the heat sink fins. Finally, the positioned heat sink assembly is transported to… The heat sink is fed into a nitrogen-protected continuous brazing furnace or a vacuum brazing furnace, where the connection between the heat sink substrate and the heat sink fins is completed according to the set heating rate, peak brazing temperature and holding time, forming a heat sink body with a continuous heat conduction path. After the brazing is completed, the heat sink body is transported to a CNC precision milling machine or a surface grinding machine to perform precision machining on the contact bottom surface of the heat sink that is used to contact the SoC chip package surface or the thermal cover plate, in order to control the flatness and surface roughness of the contact bottom surface. Then, an ultrasonic cleaner or a spray cleaning line is used to remove machining chips, residual flux and surface contaminants, and the heat sink is sent to an anodizing treatment line or an insulating coating equipment for surface protection treatment according to the protection requirements.

[0079] During the finished product inspection stage, an image measuring instrument or coordinate measuring machine is used to inspect the external dimensions, mounting hole positions, and fin arrangement of the heat sink substrate. A laser displacement measuring instrument or flatness inspection platform is used to inspect the flatness of the chip contact surface, and a surface roughness meter is used to inspect the roughness of the contact surface. In the thermal performance verification stage, an automatic thermal grease dispensing machine or thermal pad attaching equipment is used to place thermally conductive interface material between the heat sink contact surface and the simulated SoC chip heat source module. A servo press-fit platform or assembly fixture with a pressure sensor is used to press the heat sink against the simulated heat source module with a preset clamping force. Subsequently, the assembled test component is connected to a thermal performance test bench. A programmable heating module simulates the heating state of the SoC chip under target power consumption. Temperature change data of the heat sink contact area and fin area are acquired using a thermocouple array, infrared thermal imager, and temperature data acquisition device. Thermal resistance, temperature rise, and heat diffusion uniformity are then calculated to determine whether the produced SoC heat sink meets the corresponding chip's heat dissipation requirements.

[0080] Based on the method of Embodiment 1, this application also provides Embodiment 2, as follows:

[0081] When a target SoC chip has multiple thermal contact areas with different thermal input intensities under target operating conditions—for example, when a target SoC chip assembled in an edge computing terminal simultaneously performs image acquisition and processing tasks, target recognition and inference tasks, and communication data upload tasks—the heat flux density experienced by the thermal contact areas corresponding to the image processing unit and neural network operation unit is higher than that experienced by the thermal contact areas corresponding to the basic control unit and communication interface unit. To avoid the temperature rise deviation corresponding to the low thermal input area from having an excessive impact on the overall evaluation results, the boundary parameters and / or candidate production process parameters corresponding to subsequent production batches are updated based on the thermal performance test results. This also includes: reading the simulated temperature rise results and simulated thermal diffusion uniformity results corresponding to the target production process parameter scheme in the thermal performance simulation model, and dividing the maximum value in the actual temperature rise test results and the maximum value in the actual thermal diffusion uniformity test results into... The maximum thermal performance deviation result is obtained by comparing it with the maximum value in the corresponding simulation results. Specifically, it includes the absolute value of the difference between the maximum actual temperature rise value and the maximum simulated temperature rise value of each bottom heated area, and the absolute value of the difference between the maximum actual temperature rise value and the maximum simulated temperature rise value of each fin connection area. It is then determined whether the maximum thermal performance deviation result meets the preset maximum allowable deviation condition. If it does, the target production process parameter scheme is determined as the confirmed production process parameter scheme for the subsequent production batch. If it does not meet the condition, the correction values ​​of the corresponding interface heat transfer boundary parameters and / or connection heat transfer boundary parameters are updated to obtain the updated boundary parameters. Based on the updated boundary parameters, the target thermal performance simulation model is updated, and the target production process parameter scheme for the subsequent production batch is regenerated.

[0082] Among them, the preset maximum allowable deviation condition means that the absolute value of the difference between the maximum actual temperature rise value corresponding to each bottom surface heated area and the maximum simulated temperature rise value corresponding to each bottom surface heated area is less than or equal to the preset maximum deviation threshold for bottom surface heating, and the absolute value of the difference between the actual maximum temperature rise value between each fin connection associated area and the simulated maximum temperature rise value between each fin connection associated area is less than or equal to the preset maximum deviation threshold for connection association.

[0083] In this embodiment, when the target SoC chip has multiple thermal contact areas with different thermal input intensities, by comparing the maximum temperature rise value in the actual temperature rise detection results and the maximum temperature rise difference in the actual thermal diffusion uniformity detection results with the maximum simulated temperature rise value and the maximum simulated temperature rise difference corresponding to the target thermal performance simulation model, the evaluation focus can be concentrated on high heat input areas and areas with uneven thermal diffusion. This processing method can more directly reflect the local hot spot control effect and heat diffusion state of the heat sink fin area of ​​the target SoC chip under high-load tasks such as image processing, target recognition inference, and communication uploading. When the maximum thermal performance deviation result meets the preset maximum allowable deviation condition, it can be confirmed that the target production process parameter scheme has good adaptability to both high heat input areas and key areas of thermal diffusion. When the condition is not met, the interface heat transfer boundary parameters and / or connection heat transfer boundary parameters are reversed according to the maximum temperature rise deviation and the maximum temperature rise difference deviation, and the target production process parameter scheme corresponding to the subsequent production batch is regenerated. This forms a closed-loop correction mechanism targeting key hot spots and main heat diffusion paths, creating a continuous correlation between simulation model updates, boundary parameter corrections, and production process parameter optimizations. This helps improve the SoC heatsink's heat dissipation adaptability to local high heat flux density areas under complex operating conditions, and enhances the consistency and reliability of heat dissipation performance in subsequent production batches.

[0084] Through the above description of the implementation methods, those skilled in the art can clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the above functions can be divided into different functional modules to complete all or part of the functions described above.

[0085] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A data-driven adaptive optimization method for heat sink manufacturing process parameters, characterized in that, Includes the following steps: The thermal input data of the target SoC chip under the target operating conditions is obtained, and the production process parameter data and production status detection data of the SoC heat sink in the current production batch are obtained. Based on the heat transfer relationship between the target SoC chip, the heat sink substrate and the heat sink fins, the thermal input data and the production status detection data are correlated to generate thermal transfer status data. The loading data of the initial thermal performance simulation model based on the SoC heatsink is corrected according to the heat transfer state data, and the target thermal performance simulation model corresponding to the actual heat transfer state of the current production batch of SoC heatsink is obtained based on the obtained boundary parameters. Based on the target thermal performance simulation model, the thermal diffusion uniformity corresponding to different candidate production process parameter schemes is simulated and evaluated, and the target production process parameter scheme is determined from the candidate production process parameter schemes according to the simulation evaluation results. The target production process parameter scheme is used for the production of subsequent batches of SoC heat sinks, and the corresponding thermal performance test results are obtained. The boundary parameters and / or the candidate production process parameter schemes corresponding to subsequent batches are updated based on the thermal performance test results.

2. The data-driven adaptive optimization method for radiator manufacturing process parameters as described in claim 1, characterized in that, The specific process for acquiring the production status monitoring data is as follows: Obtain the target running task, data processing load, and task duration corresponding to the target SoC chip, and combine them to define the running condition data under the target running condition; The data processing load refers to the amount of data processing work undertaken by the target SoC chip during the execution of the corresponding target running task, and can be determined by at least one of processor core utilization, memory access frequency, interface data transmission rate and / or operating power consumption; The task duration refers to the length of time during which the target SoC chip continuously executes the target task according to the corresponding data processing load. The target SoC chip refers to the system-on-chip chip that is to be adapted for heat dissipation with the SoC heat sink of the current production batch, and serves as the target heat source object for generating thermal input data. The target operating condition refers to the working state in which the target SoC chip to be adapted performs a set combination of functional tasks in a predetermined application device and continuously runs according to the corresponding data processing load level; Under the target operating conditions, based on the pre-built radiator thermal matching model, the operating condition data is processed by inputting the radiator thermal matching model to obtain the heat input intensity, heat input duration and heat input distribution location corresponding to each heat contact area, which are then used as the heat input data. Obtain the batch identifier and sample identifier corresponding to each sample SoC heatsink in the current production batch, and read the initial production process parameters of the corresponding sample SoC heatsink from the historical production equipment control record based on the batch identifier and sample identifier. The connection status of the heat sink substrate and heat sink fin connection area of ​​each SoC heat sink sample in the current production batch is detected by industrial vision inspection equipment to obtain connection formation status data. Based on industrial vision inspection equipment and three-dimensional contour measurement equipment, the spreading state of the thermal interface material disposed on the contact bottom surface of the heat sink is detected, and the spreading state data of the thermal interface material is obtained. The connection forming state data and thermal interface material spreading state data corresponding to the same sample identifier are combined to form the production state detection data, and the production state detection data is associated with and stored with the initial production process parameters corresponding to the same sample identifier.

3. The data-driven adaptive optimization method for radiator manufacturing process parameters as described in claim 2, characterized in that, The process of associating the thermal input data with the production status detection data based on the heat transfer relationship between the target SoC chip, the heat sink substrate, and the heat sink fins is as follows: Based on the preset assembly positioning relationship between the target SoC chip package surface and the SoC heat sink contact bottom surface, the bottom surface heating area corresponding to each thermal contact area on the heat sink contact bottom surface is obtained. The preset assembly positioning relationship indicates the spatial positional correspondence between the target SoC chip package surface or the corresponding thermal conductive cover and the contact bottom surface of the SoC heat sink in a predetermined assembly state. Based on the heat input distribution location corresponding to each heat contact area, the corresponding heat input intensity and heat input duration are bound to the bottom heated area to obtain the regional heat input state corresponding to each bottom heated area. The specific process is as follows: Read the region number, planar coordinate range and region boundary range corresponding to any heat contact area, and record the heat input intensity and heat input duration corresponding to the heat contact area to the corresponding bottom heated area, so that each bottom heated area has corresponding heat input state data, including the heat flux density and heat flux density duration of the bottom heated area.

4. The data-driven adaptive optimization method for radiator manufacturing process parameters as described in claim 3, characterized in that, The step of associating the thermal input data with the production status detection data further includes: Based on the actual area covered by the thermal interface material in each thermal contact area in the thermal interface material spreading state data, the effective coverage range of the interface corresponding to each bottom heated area is determined, and the effective coverage range of the interface is associated with the corresponding area thermal input state to obtain the interface heat transfer state between the target SoC chip packaging surface and the bottom contact surface of the heat dissipation substrate. When the area actually covered by the heat-conducting interface material in the bottom heated area is smaller than the area of ​​the corresponding bottom heated area, the uncovered area is determined as the heat transfer missing area in the corresponding interface heat transfer state. When the area actually covered by the heat-conducting interface material in the bottom heated area is greater than or equal to the area of ​​the corresponding bottom heated area, the bottom heated area is determined as the interface continuous coverage area. The effective coverage area of ​​the interface is associated with the corresponding regional heat input state. For the same bottom heated area, the heat flux density and heat flux density duration corresponding to the bottom heated area are bound to the effective coverage area of ​​the interface. Obtain the substrate outline, heat dissipation fin arrangement position, and connection area position between the heat dissipation substrate and the heat dissipation fins corresponding to the heat dissipation substrate. Take each bottom heated area as the heat input starting area. According to the heat diffusion direction extending from the contact bottom to the heat dissipation fin connection area inside the heat dissipation substrate, determine the substrate diffusion association area and the corresponding fin connection association area corresponding to each bottom heated area. Based on the connection formation status data, the connection continuity status corresponding to each fin connection association area is determined. The specific process is as follows: Read the effective continuous connection length of the fin connection association area corresponding to the connection formation status data, compare the effective continuous connection length with the preset continuous connection length threshold of the corresponding fin connection association area to obtain the connection continuity status. When the effective continuous connection length of the fin connection association area is greater than or equal to the preset continuous connection length threshold, the corresponding fin connection association area is determined to be in a connection continuity satisfied state. Otherwise, the corresponding fin connection association area is determined to be in a connection continuity insufficient state, and the corresponding connection interruption position and connection interruption length are recorded. The continuous connection state is bound to the substrate diffusion association region corresponding to the bottom heated area to obtain the connection heat transfer state corresponding to the heat transfer substrate to the heat transfer fins. For each bottom heated area, the corresponding regional heat input state, interface heat transfer state, substrate diffusion associated area, and connection heat transfer state are combined to generate corresponding heat transfer path state data. The heat transfer path state data corresponding to multiple bottom heated areas are collected to generate heat transfer state data.

5. The data-driven adaptive optimization method for radiator manufacturing process parameters as described in claim 4, characterized in that, The specific process for obtaining the target thermal performance simulation model corresponding to the actual heat transfer state of the current production batch of SoC heat sinks is as follows: Read the heat transfer state data, and generate corresponding interface heat transfer boundary correction parameters based on the continuous coverage area or heat transfer gap location within each bottom surface heated area, specifically: When a continuous coverage area of ​​the interface is detected, the correction value of the interface heat transfer boundary parameter is not obtained, and the initial interface heat transfer boundary parameter of the corresponding interface region in the initial thermal performance simulation model is kept as the set interface heat transfer boundary parameter. When a location with a lack of heat transfer is detected, the data on the spread state of the thermally conductive interface material at that location is substituted into the pre-constructed correction model of the interface heat transfer boundary parameters to obtain the correction value of the interface heat transfer boundary parameters, which is then used as the set interface heat transfer boundary parameters.

6. The data-driven adaptive optimization method for radiator manufacturing process parameters as described in claim 5, characterized in that, The method of obtaining the target thermal performance simulation model corresponding to the actual heat transfer state of the SoC heatsink in the current production batch also includes: The heat transfer state data is read, and based on the continuity of connection or the insufficiency of connection corresponding to each fin connection associated region, correction values ​​for the corresponding connection heat transfer boundary parameters are generated, specifically: When the connection is determined to be in a continuous state, the correction value of the connection heat transfer boundary parameter is not obtained. The initial connection heat transfer boundary parameter of the corresponding connection area in the initial thermal performance simulation model is maintained and used as the set connection heat transfer boundary parameter. When the connection is determined to be in a continuous state, the connection interruption position and connection interruption length corresponding to the fin connection associated area are input into the correction model of the connection heat transfer boundary parameter. The correction value of the connection heat transfer boundary parameter output by the model is used as the set connection heat transfer boundary parameter. The boundary parameters and corresponding thermal input data are loaded into the initial thermal performance simulation model to obtain the target thermal performance simulation model corresponding to the actual heat transfer state of the SoC heat sink in the current production batch.

7. The data-driven adaptive optimization method for radiator manufacturing process parameters as described in claim 6, characterized in that, The specific process for determining the target production process parameter scheme from the candidate production process parameter schemes based on the simulation evaluation results is as follows: Read the simulation results of temperature rise distribution and heat diffusion uniformity of the current production batch of SoC heat sinks output by the target thermal performance simulation model; The simulation results include the temperature rise values ​​corresponding to each heated area on the bottom surface and the temperature rise values ​​corresponding to each fin connection and related area. Based on the simulation results, the target heat transfer path with heat dissipation performance deviation is determined. When the temperature rise value corresponding to any bottom heated area is greater than the preset allowable temperature rise threshold, the heat transfer path containing the bottom heated area and its corresponding interface heat transfer state is determined as the interface heat transfer deviation path. Conversely, the heat transfer path containing the bottom heated area and its corresponding interface heat transfer state is determined as the interface heat transfer normal path. When the temperature rise value corresponding to any fin connection associated area is greater than the preset connection associated allowable temperature rise threshold, the heat transfer path containing the fin connection associated area and its corresponding connection heat transfer state is determined as the connection heat transfer deviation path; otherwise, the heat transfer path containing the corresponding fin connection associated area and its corresponding connection heat transfer state is determined as the connection heat transfer normal path. At least one of the interface heat transfer deviation path and the connection heat transfer deviation path is determined as the target heat transfer path.

8. The data-driven adaptive optimization method for radiator manufacturing process parameters as described in claim 7, characterized in that, The scheme for determining the target production process parameters also includes: Based on the interface heat transfer state and / or connection heat transfer state corresponding to the target heat transfer path, the production process parameters to be adjusted are determined, and the specific process is as follows: The simulation results corresponding to the target heat transfer path and the initial production process parameters are input into the heat sink production process parameter readjustment model based on machine learning to determine the production process parameters to be adjusted. Based on the current parameter values ​​corresponding to the production process parameters to be adjusted, a candidate production process parameter scheme is generated. Based on the candidate production process parameter scheme, the simulation results of the temperature rise distribution and heat diffusion uniformity corresponding to the current production batch of SoC heat sinks output by the target thermal performance simulation model are reread. When the simulation results meet the preset diffusion uniformity conditions, the candidate production process parameter scheme is determined as the target production process parameter scheme; otherwise, a simulation evaluation alarm is sent.

9. The data-driven adaptive optimization method for radiator manufacturing process parameters as described in claim 1, characterized in that, The specific process for updating the boundary parameters and / or candidate production process parameters for subsequent production batches based on the thermal performance test results is as follows: The target production process parameter scheme is sent to the corresponding production equipment so that subsequent batches of SoC heat sinks are produced in accordance with the target production process parameter scheme. Obtain a sample SoC heatsink produced according to the target production process parameter scheme, and perform thermal performance testing on it to obtain the corresponding actual temperature rise test results and actual thermal diffusion uniformity test results; Read the simulation temperature rise result and simulation heat diffusion uniformity result corresponding to the target production process parameter scheme in the target thermal performance simulation model, and compare the actual temperature rise detection result and the actual heat diffusion uniformity detection result with the corresponding simulation results to obtain the thermal performance deviation result; Determine whether the thermal performance deviation result meets the preset allowable deviation condition. If it does, then the target production process parameter scheme is determined as the confirmed production process parameter scheme for the subsequent production batch. If it does not meet the condition, then the correction values ​​of the corresponding interface heat transfer boundary parameters and / or connection heat transfer boundary parameters are updated to obtain the updated boundary parameters. Based on the updated boundary parameters, the target thermal performance simulation model is updated, and the target production process parameter scheme for the subsequent production batch is regenerated.

10. The data-driven adaptive optimization method for radiator manufacturing process parameters as described in claim 1, characterized in that, The step of updating the boundary parameters and / or candidate production process parameter schemes corresponding to subsequent production batches based on the thermal performance test results further includes: Read the simulated temperature rise result and simulated heat diffusion uniformity result of the target production process parameter scheme in the thermal performance simulation model, and compare the maximum value in the actual temperature rise detection result and the maximum value in the actual heat diffusion uniformity detection result with the maximum value in the corresponding simulation result to obtain the maximum thermal performance deviation result; Determine whether the maximum thermal performance deviation result meets the preset maximum allowable deviation condition. If it does, then the target production process parameter scheme is determined as the confirmed production process parameter scheme for the subsequent production batch. If it does not meet the condition, then the correction values ​​of the corresponding interface heat transfer boundary parameters and / or connection heat transfer boundary parameters are updated to obtain the updated boundary parameters. Based on the updated boundary parameters, the target thermal performance simulation model is updated, and the target production process parameter scheme for the subsequent production batch is regenerated.

Images