A spectral and color-separated BIN-level linked intelligent control system and method for the entire process of LED manufacturing

The LED full-process intelligent control system, which links light and color separation with BIN level, solves the problem of the inability to quantify the time-series thermal accumulation effect of the entire LED production process in existing technologies. It realizes the linkage quantitative control of all process parameters and dynamic BIN level, ensuring precise quality control in LED production.

CN123113549APending Publication Date: 2026-11-17广州市昊链信息科技股份有限公司
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611402720.X
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-10
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In the current LED production process, the instantaneous single-point parameter acquisition and fixed threshold judgment mode cannot quantify the time-series thermal accumulation effect of the entire process, resulting in continuous cumulative damage from heat loss and structural displacement. This makes it impossible to achieve accurate BIN grading and batch quality consistency, leading to misjudgment of qualified products and omission of unqualified products, which makes it difficult to meet the quality control requirements of high-precision LED mass production.

Method used

The system adopts a spectral and color-separated BIN-level linkage LED full-process intelligent control system. Through the thermal accumulation-structure coupling calculation module, dynamic hierarchical anomaly judgment module, and full-process linkage handling and traceability module, it realizes the collection and linkage calculation of multi-dimensional time series data, dynamically adjusts the BIN judgment threshold, locates the source of process anomalies, and performs precise control.

Benefits of technology

It achieves coordinated and quantitative control of all process parameters, accurately quantifies accumulated process damage, dynamically adapts to BIN grading standards, accurately locates the source of anomalies, avoids misjudgment and omission of products, and ensures the accuracy of BIN grading of finished LED products and the consistency of batch production.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123113549A_ABST
    Figure CN123113549A_ABST
Patent Text Reader

Abstract

This invention belongs to the field of LED production quality control technology, and discloses an intelligent control system and method for LED full-process linkage of spectral and color-separation BIN level. The system includes a thermal accumulation-structure coupling calculation module, a dynamic graded anomaly judgment module, and a full-process linkage handling and traceability module, with each module connected and cooperating in sequence. This invention collects time-series thermal and structural parameters of the entire LED process, calculates the thermal accumulation impact by integration, obtains the coupling damage coefficient by coupling process structural parameters, corrects the photoelectric parameter judgment standard based on the damage coefficient, and outputs the BIN offset judgment coefficient. Based on the coefficient range, the anomaly level is defined and corresponding control is executed, tracing the source of the process. This invention abandons the defects of traditional instantaneous single-point judgment, independent parameter judgment, and fixed threshold judgment, and realizes the linkage quantitative control of parameters throughout the entire LED production process, ensuring the accuracy of LED spectral and color-separation BIN grading and the consistency of production quality.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of LED manufacturing process quality control technology, specifically to an intelligent control system and method for the entire process of LED manufacturing with BIN-level linkage for spectral and color separation. Background Technology

[0002] Currently, the industry standard for quality control in LED production employs instantaneous single-point parameter acquisition and fixed threshold judgment. This approach only samples and compares photoelectric and temperature parameters at a single moment during die bonding, wire bonding, packaging, and color sorting, without quantifying the cumulative thermal effects throughout the entire production process. Heat loss and structural deviations at each stage of LED production generate continuous cumulative damage. Instantaneous sampling data cannot characterize the cumulative process defects resulting from continuous production. Furthermore, existing technologies independently judge thermal parameters, process structural parameters, and photoelectric parameters, lacking inter-parameter coupling logic. This fails to reflect the correlation between process damage and the accuracy of LED color sorting and binning (BIN) grading. Fixed BIN judgment thresholds cannot adapt to LED products with varying degrees of process damage, easily leading to misjudgments of qualified products and omissions of unqualified products. This results in chaotic BIN grading of finished LEDs, poor batch-to-batch quality consistency, and an inability to achieve full-process traceability and precise control, ultimately failing to meet the quality control requirements of high-precision LED mass production. Summary of the Invention

[0003] The purpose of this invention is to overcome the shortcomings of the existing technology and propose a full-process intelligent control system for LEDs with spectral and color separation BIN level linkage, including a thermal accumulation-structure coupling calculation module, a dynamic hierarchical anomaly judgment module, and a full-process linkage handling and traceability module.

[0004] The thermal accumulation-structure coupling calculation module is used to receive multi-dimensional time-series acquisition data corresponding to the entire process of LED die bonding, wire bonding, packaging, and color sorting; calculate the thermal accumulation impact based on the thermal-related parameters inside the multi-dimensional time-series acquisition data; and calculate the coupling damage coefficient based on the thermal accumulation impact and the structural parameters inside the multi-dimensional time-series acquisition data.

[0005] The dynamic hierarchical anomaly determination module is communicatively connected to the thermal accumulation-structure coupling calculation module. It is used to receive the coupling damage coefficient, calculate the BIN offset determination coefficient based on the coupling damage coefficient and the measured value of LED photoelectric parameters, and lock the production anomaly level according to the numerical range corresponding to the BIN offset determination coefficient.

[0006] The full-process linkage handling and traceability module is communicatively connected to the dynamic hierarchical anomaly judgment module. It is used to receive the production anomaly level, trigger production control operations that match the production anomaly level, and locate the process source type that caused the production anomaly.

[0007] Preferably, it also includes a product parameter dynamic configuration and switching module, wherein the product parameter dynamic configuration and switching module has a built-in parameter adaptation database;

[0008] The product parameter dynamic configuration and model change module is used to receive the target LED product model parameters entered externally, retrieve the thermal accumulation influence weight coefficient, structural offset weight coefficient, contact length weight coefficient, maximum allowable thermal accumulation threshold, and maximum process offset threshold that match the target LED product model parameters in the parameter matching database, and synchronously transmit all matching parameters to the thermal accumulation-structure coupling calculation module and the dynamic hierarchical anomaly judgment module.

[0009] Further preferably, it also includes a full-process multi-dimensional time-series acquisition module, which is communicatively connected to the thermal accumulation-structure coupling calculation module and the product parameter dynamic configuration and change module, respectively. The full-process multi-dimensional time-series acquisition module includes a time-series thermal data acquisition sub-module, a process structure data acquisition sub-module, and a photoelectric parameter acquisition sub-module.

[0010] The time-series thermal data acquisition submodule is used to collect process temperature data and equipment power loss data throughout the entire LED production process.

[0011] The process structure data acquisition submodule is used to acquire the offset amount of the LED die bonding process, the offset amount of the wire bonding process, and the effective contact length of the LED manufacturing process.

[0012] The photoelectric parameter acquisition submodule is used to acquire the measured values ​​of the forward voltage and luminous flux of the LED device.

[0013] The full-process multi-dimensional time-series acquisition module binds a unique product management ID to all acquired data, generates full-process multi-dimensional time-series acquisition data, and completes data output.

[0014] More preferably, the thermal accumulation-structure coupling calculation module is configured with progressive linkage calculation logic, which includes time-series thermal integral calculation logic, multi-parameter normalization coupling calculation logic, and photoelectric parameter correction input logic.

[0015] The time-series heat integral calculation logic is used to perform time integral calculations on the time-series heat-related parameters of the entire process and output the corresponding cumulative heat influence.

[0016] The multi-parameter normalized coupling calculation logic is used to receive the thermal cumulative influence, the process structure offset, and the effective contact length of the process, to complete the multi-dimensional process parameter normalized coupling calculation, and to output the corresponding coupling damage coefficient.

[0017] The photoelectric parameter correction input logic is used to transmit the coupling damage coefficient to the dynamic hierarchical anomaly determination module in a directional manner, providing a fixed input variable for the photoelectric parameter offset correction calculation.

[0018] More preferably, the time-series thermal integration calculation logic is configured with process time-series thermal accumulation quantization operation rules, the multi-parameter normalization coupling calculation logic is configured with process damage quantization operation rules, and the photoelectric parameter correction input logic is configured with BIN level offset quantization operation rules.

[0019] The calculation dimensions of the process timing thermal accumulation quantification calculation rules, process damage quantification calculation rules, and BIN level offset quantification calculation rules are mutually compatible. All calculation rules rely on the LED full-process production parameters to establish a normalized linkage calculation system, realizing the linkage quantification processing of thermal parameters, structural parameters, and photoelectric parameters.

[0020] More preferably, the dynamic hierarchical anomaly determination module has a built-in fixed BIN hierarchical determination standard table. The BIN hierarchical determination standard table is divided into three independent numerical intervals. The three independent numerical intervals correspond one-to-one with the numerical values ​​of the BIN offset determination coefficient. The three independent numerical intervals are respectively matched with three anomaly levels: warning level anomaly, rework level anomaly, and shutdown level anomaly.

[0021] More preferably, the full-process linkage handling and traceability module has a built-in three-level independent control logic, which corresponds one-to-one with the early warning level anomaly, the rework level anomaly, and the shutdown level anomaly;

[0022] The three independent control logics are used to push out instructions for fine-tuning process parameters, locking and rework of product batches, and shutdown and process rectification of production lines, respectively, to complete control operations such as correcting minor process deviations, intercepting non-conforming products, and shutting down and rectifying production line malfunctions.

[0023] More preferably, the full-process linkage handling and traceability module has a built-in fixed source classification logic. The source classification logic reads the multi-dimensional time-series data collected from the entire process and the coupling damage coefficient, and matches and determines the three types of process anomaly sources: thermal accumulation overload type, process structure deviation type, and photoelectric parameter drift type.

[0024] The full-process linkage and traceability module integrates anomaly level data and process source type data to generate a standardized electronic traceability report and completes storage and archiving.

[0025] A method for intelligent control of the entire process of LEDs with spectral and color separation BIN-level linkage, characterized in that the intelligent control system for the entire process of LEDs with spectral and color separation BIN-level linkage, applied as described above, is executed by an industrial measurement and control processor and includes the following steps:

[0026] S1. Enter the model parameters of the target LED product, retrieve the matching weight coefficients and process threshold parameters, and complete the adaptation configuration of the system control model.

[0027] S2. Collect thermal, structural, and optoelectronic parameters of the entire LED manufacturing process, including die bonding, wire bonding, packaging, and color sorting, and construct a multi-dimensional time-series dataset bound to the product management ID;

[0028] S3. Solve the cumulative thermal impact by integrating the thermal parameters of the multidimensional time series dataset, and determine the thermal damage status of the process by combining the preset threshold.

[0029] S4. Integrate the cumulative thermal impact, process structure offset data, and effective contact length data to calculate the coupling damage coefficient;

[0030] S5. Based on the coupling damage coefficient, the photoelectric parameter judgment standard is corrected, the BIN offset judgment coefficient is calculated, and the abnormality level is determined by matching the numerical range.

[0031] S6. Perform corresponding process fine-tuning, batch rework, and production line shutdown rectification operations according to the abnormality level;

[0032] S7. Retest the parameters of abnormal batches of products and verify the process status, locate the source of process abnormality, and iteratively update the system process thresholds and weight parameters.

[0033] In a further preferred embodiment, steps S3, S4, and S5 form a progressively advancing closed-loop calculation process. The quantified data output from the preceding steps serves as the sole core input for the subsequent steps, enabling the entire process to achieve a closed-loop linkage determination of thermal parameters, structural parameters, and photoelectric parameters for the entire LED process, and completing the synchronous execution of process damage quantification and BIN grading dynamic correction.

[0034] Technical effects:

[0035] This invention addresses the core shortcomings of existing LED control methods, such as instantaneous sampling, fragmented parameters, and fixed thresholds. It achieves comprehensive quantitative control of all process parameters through a linked architecture that quantifies the cumulative impact of thermal processes via time-series integration, calculates damage coefficients by coupling multiple process parameters, and dynamically adjusts the BIN (Browser Ingress) determination threshold. This invention integrates the logical relationships between thermal, structural, and optoelectronic parameters, accurately quantifies accumulated process damage, dynamically adapts BIN grading standards to different damage states, precisely distinguishes multiple levels of production anomalies and matches corresponding control strategies, and accurately locates the source of process anomalies. This effectively avoids misjudgments and omissions, ensuring the accuracy of BIN grading in finished LED products and consistency with batch production. Attached Figure Description

[0036] Figure 1 This is a connection block diagram of the intelligent control system for the entire LED process with BIN-level linkage for spectral and color separation, as described in this invention.

[0037] Figure 2 This is a flowchart of the intelligent control method for the entire process of LEDs with BIN-level linkage for spectral and color separation, as described in this invention. Detailed Implementation

[0038] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0039] Please see Figures 1-2 Current LED production quality control methods mostly rely on instantaneous single-point sampling to obtain production parameters, which fails to capture cumulative process damage. Multiple production parameters are judged independently, with no correlation between them. The fixed BIN (Body Ingress) grading threshold cannot adapt to the product assessment requirements of different process damage states. These factors collectively lead to grading errors and quality control failures.

[0040] Based on this, this embodiment provides an intelligent control system for the entire LED process, with BIN-level linkage for spectral and color separation. Its hardware and logic architecture consists of a thermal accumulation-structure coupling calculation module, a dynamic hierarchical anomaly determination module, and a full-process linkage handling and traceability module. These three modules form a fixed signal transmission and computation closed loop. The thermal accumulation-structure coupling calculation module receives multi-dimensional time-series acquisition data corresponding to the entire LED process, including die bonding, wire bonding, packaging, and spectral and color separation. Based on the thermally relevant parameters within this data, it calculates the thermal accumulation impact, and then calculates the coupling damage coefficient based on the thermal accumulation impact and the structural parameters within the multi-dimensional time-series acquisition data. The dynamic hierarchical anomaly determination module communicates with the thermal accumulation-structure coupling calculation module, receives the coupling damage coefficient, calculates the BIN offset determination coefficient based on the coupling damage coefficient and the measured values ​​of the LED photoelectric parameters, and locks the production anomaly level according to the numerical range within which the coefficient falls. The full-process linkage handling and traceability module communicates with the dynamic hierarchical anomaly determination module, receives the production anomaly level, triggers production control operations matching the level, and simultaneously locates the process source type that caused the production anomaly. The thermal accumulation-structure coupling calculation module performs integral calculations based on full-time acquired data to quantify the thermal accumulation damage throughout the LED production process. It also integrates structural offset and contact length parameters from die bonding and wire bonding processes, transforming accumulated process damage into a quantifiable coupling damage coefficient. The dynamic grading anomaly judgment module uses process damage data as a correction basis, enabling the BIN grading standard to dynamically adapt to the process status. The full-process linkage handling and traceability module matches the judgment results with production control actions and completes anomaly tracing, forming a closed-loop control system from data acquisition, calculation, judgment, control to tracing.

[0041] The system also includes a dynamic configuration and model change module for product parameters, which has a built-in parameter adaptation database. When the target LED product model parameters are entered into the module from an external source, the module retrieves the thermal accumulation influence weight coefficient, structural offset weight coefficient, contact length weight coefficient, maximum allowable thermal accumulation threshold, and maximum process offset threshold that match that model from the database, and synchronously transmits these parameters to the thermal accumulation-structure coupling calculation module and the dynamic grading anomaly judgment module. The database adopts a classified storage structure, with each product model corresponding to a set of independent parameter records, and each field in the record corresponds one-to-one with the parameters required by the calculation model. When the production line changes models, the operator only needs to enter or select the product model code on the interactive interface, and the module automatically completes the retrieval and distribution. All relevant variables in the calculation and judgment modules are updated in real time, without the need for manual setting. The parameter synchronization transmission mechanism is completed through shared memory or industrial bus, ensuring that the parameters referenced in thermal accumulation calculation, coupling damage calculation, and BIN grading judgment originate from the same version, avoiding control errors caused by parameter inconsistencies.

[0042] To provide a complete data foundation for the computing module, the system also includes a multi-dimensional time-series acquisition module covering all processes. This module communicates with the thermal accumulation-structure coupling calculation module and the product parameter dynamic configuration and changeover module. Internally, the multi-dimensional time-series acquisition module is divided into a time-series thermal data acquisition submodule, a process structure data acquisition submodule, and a photoelectric parameter acquisition submodule. The time-series thermal data acquisition submodule continuously acquires temperature data near the LED devices within the equipment and real-time power loss data of the equipment drive system at a fixed sampling period during the complete execution time of each process—LED die bonding, wire bonding, packaging, and color sorting. Temperature data is acquired using armored thermocouples or infrared temperature sensors placed near the heat source of the equipment, while power loss data is acquired by the electrical parameter measurement module, which collects the input voltage and current of the equipment in real time and calculates the active power loss. After the die bonding process, the process structure data acquisition submodule uses a vision positioning system to capture images of the relative position of the chip and the substrate. The chip offset is calculated using an image processing algorithm and recorded in micrometers. The same method is used to acquire the wire bonding offset after the wire bonding process. The effective contact length is obtained by measuring the geometric dimensions of the area covered by the die bonding solder or thermally conductive adhesive. In the spectral separation process, the photoelectric parameter acquisition submodule uses an integrating sphere and a spectrometer to measure the measured forward voltage and luminous flux of the LED device. The full-process multi-dimensional timing acquisition module assigns a unique product control ID to each set of acquired data. This ID is used throughout the entire LED production process. All temperature sequences, power sequences, structural offsets, and photoelectric measured values ​​associated with the same LED or batch are organized using this ID as an index, generating structured full-process multi-dimensional timing acquisition data. This data is then output in real-time to the thermal accumulation-structure coupling calculation module via industrial Ethernet or fieldbus.

[0043] The thermal accumulation-structure coupling calculation module is internally configured with a progressive, interconnected calculation logic, consisting of three sub-logic: a time-series thermal integration calculation logic, a multi-parameter normalization coupling calculation logic, and a photoelectric parameter correction input logic. The time-series thermal integration calculation logic receives the temperature and power loss sequences from the multi-dimensional time-series data acquired throughout the entire process, performs time integration, and outputs the thermal accumulation impact. The multi-parameter normalization coupling calculation logic receives the thermal accumulation impact, the offset from the process structure data, and the effective contact length, completes the normalization coupling calculation of multi-dimensional process parameters, and outputs the coupling damage coefficient. The photoelectric parameter correction input logic uses the coupling damage coefficient as a fixed input variable and transmits it directionally to the dynamic hierarchical anomaly judgment module, providing a basis for subsequent photoelectric parameter offset correction calculations. These three layers of logic are interconnected, with the output value of the previous layer becoming the input of the next layer, forming a progressive calculation chain. To ensure the standardization and consistency of the calculations, the time-series thermal integral calculation logic, the multi-parameter normalization coupling calculation logic, and the photoelectric parameter correction input logic are respectively configured with process time-series thermal accumulation quantization calculation rules, process damage quantization calculation rules, and BIN level offset quantization calculation rules. The calculation dimensions of these three sets of rules have been uniformly adapted and calibrated to ensure that the calculation results of thermal parameters, structural parameters, and photoelectric parameters are consistent in dimensions when coupled, and can be superimposed and reused.

[0044] Under the process-sequence heat accumulation quantization calculation rules, the time-sequence heat integral calculation logic is executed based on the multi-process time-sequence superposition heat accumulation coefficient calculation model. The calculation form of this model is as follows: This model is based on the thermodynamic law of energy accumulation and the principle of heat loss due to power consumption in industrial equipment. During LED production, processes such as die bonding, wire bonding, and packaging require applying certain temperatures and pressures to the device. The heat field and heat dissipation generated by the equipment operation continuously act on the device. High instantaneous temperature does not necessarily mean large cumulative thermal damage; cumulative damage depends on the time integral of the thermal effect over the entire process duration. Therefore, the model adopts a definite integral form, representing the production time of a single process. to By continuously summing the integrand over the integration interval, the coefficients characterizing the total accumulated heat energy of the process are obtained. Its dimension is joule. The integrand is the sum of the temperature shift contribution term and the power loss contribution term. In the temperature shift contribution term, It is a moment The measured temperature inside the equipment is the difference between the actual temperature and the standard process temperature, expressed in Kelvin. The standard process temperature is set according to the ideal process temperature for each process of this LED model, while the actual temperature is obtained through online detection by a temperature sensor. The larger the value, the further the process temperature deviates from the ideal state at that moment, and the stronger the resulting thermal stress damage. (Coefficient) It is the temperature influence weighting coefficient, with the dimension of watts per Kelvin. This coefficient converts the pure temperature quantity of temperature difference into the thermal damage rate expressed in power equivalent, so that this term can be unified with the power loss term in terms of dimension. The determination method can be achieved by conducting thermal stress aging tests on a specific LED model under different temperature offset conditions, measuring the junction temperature change and performance degradation rate of the device, calculating the ratio of temperature stress to equivalent damage power using the Arrhenius relation, and then obtaining the result through linear regression. In the power loss contribution term, Is the equipment in constant The real-time operating power loss is measured in watts. In LED production equipment such as wire bonders and die bonders, a portion of the actual power consumption of ultrasonic generators or heating platforms is not directly used for effective processes, but is dissipated as heat. This heat acts on LED devices through conduction and radiation, becoming another source of thermal damage. The values ​​are obtained through real-time electrical parameter acquisition. Coefficient It is a dimensionless weighting coefficient for the influence of power loss, and its function is to adjust the conversion ratio between the heat dissipation and the actual heat damage absorbed by the device. The heat transfer efficiency model can be established and calibrated by comparing synchronous monitoring data of device power consumption and device temperature rise. In actual integral calculations, the time-series heat integration calculation logic employs a numerical integration method, discretizing the continuous time into... Each sampling interval, for each interval within and The function values ​​are calculated, summed, and multiplied by the interval length to obtain an approximate integral value. This model transforms the originally scattered instantaneous temperature and power data into a unified energy total value, which is the cumulative thermal influence, providing a numerical basis for characterizing the thermal history of the process in subsequent coupled calculations.

[0045] Under the multi-parameter normalized coupled calculation logic, the process damage quantification calculation rules are executed based on the multi-dimensional process coupled structure damage coefficient calculation model. The calculation form of this model is as follows: This model aims to address the challenge of directly comparing and synthesizing thermal parameters, structural offset parameters, and contact length parameters due to their varying physical properties, different dimensions, and incompatibility. The model output... The thermal coupling structural damage coefficient is dimensionless and a pure numerical value. Its magnitude represents the comprehensive damage degree resulting from the superposition of process defects in three dimensions: thermal accumulation, structural displacement, and contact length. The model consists of the sum of three weighted terms, each of which has been normalized to eliminate dimensional and scale differences in the original physical quantities. The first term is the thermal accumulation damage term. It is the cumulative heat effect calculated from the previous rule, and its dimension is joule; It is the maximum heat accumulation threshold allowed by the process standard, and its dimension is also joule. This refers to the maximum accumulated thermal energy that an LED device can withstand without affecting the final optoelectronic performance pass rate under the specified process parameters. Its value is determined through process verification tests on the same model of product. For example, by applying different accumulated thermal shocks to multiple batches of products, statistically analyzing the BIN (Browser Intake) grading pass rate, and determining the accumulated thermal energy corresponding to the pass rate threshold as the threshold value. (Ratio) The normalization of heat accumulation was achieved when Less than When this value is less than 1, it indicates that the heat damage is within the permissible range; when it exceeds 1, it means that the heat accumulation has exceeded the limit. (Coefficient) The first term is the dimensionless weight of the thermal accumulation term, reflecting the importance of thermal damage in the comprehensive assessment. The second term is the structural displacement damage term. It is the real-time process structure offset, which can be selected as die bonding offset or wire bonding offset, or the larger of the two, in meters; This is the maximum allowable offset according to process standards, also measured in meters. This value originates from product design tolerances, such as the maximum allowable positional deviation for chip mounting. (Ratio) Convert the offset into the degree of exceeding the maximum allowable value. Coefficient This is the weighting coefficient for this item. The third item is the contact length process damage item. The effective contact length is calculated from the measured solder creep height or coverage area of ​​the die bond layer, and the unit is meters; This refers to the effective contact length specified in the process standard, also in meters. Unlike the direct ratio treatment of the previous two items, the comparison value here... Square root calculations were performed. This design is based on the fact that slight deviations in effective contact length typically do not lead to a proportional degradation in thermal resistance or mechanical strength; their effects exhibit non-linear tolerance. Using a square root function better simulates this actual process characteristic, preventing the excessive amplification of minute deviations. (Coefficients) The weights for the contact length term. Three weighting coefficients. , , The constraint that the sum is 1 is satisfied, and this constraint ensures the comprehensive damage coefficient. The controllable range of values ​​allows for the production of different batches and models of products. The values ​​have comparable benchmarks. The specific values ​​of the weighting coefficients can be determined using the Taguchi method or response surface methodology to design multi-factor experiments. By analyzing the pass rate data of samples under different combinations of heat, offset, and contact length levels, variance analysis is used to decompose the contribution of each factor and its interaction, and the weights are obtained after normalization. This model integrates dispersed and heterogeneous process parameters into a single dimensionless index through normalization, nonlinear adjustment, and weighted summation, allowing the process status of the entire procedure to be calibrated using a single numerical value.

[0046] Under the input logic of photoelectric parameter correction, the BIN level offset quantization calculation rule is executed according to the dynamic determination model of BIN photoelectric parameters based on process damage coupling, and its calculation form is as follows: The purpose of this model is to generate a BIN offset determination coefficient. This serves as the final basis for BIN grading and anomaly determination of LED products. Traditional BIN grading directly compares the deviation of measured photoelectric parameters with a fixed standard threshold. This model, however, uses the coupling damage coefficient obtained in the previous step. Embedded as a dynamic correction factor in the calculation process, the grading threshold can adaptively change according to the actual process damage state. The model consists of the sum of a voltage offset correction term and a luminous flux offset correction term. In the voltage offset correction term, This is the measured value of the forward voltage. This is the standard forward voltage threshold for the target BIN level of the product, and the unit is volts. The calculated voltage relative deviation rate is dimensionless and can be positive or negative. A positive value indicates that the measured voltage is higher than the standard, and a negative value indicates that it is lower than the standard. This deviation rate is then multiplied by a correction factor. , is the base of the natural logarithm. This is the coupling damage coefficient. When the process damage coefficient... A value of 0 indicates no process damage. The voltage offset term contributes completely, which is consistent with the fact that voltage offset directly reflects the fluctuation of chip electrical characteristics in the undamaged state. With As the value increases, As the voltage offset gradually decreases and approaches zero, its contribution is compressed. The physical basis for this is that when an LED suffers severe process damage, internal chip defects increase, and the change in its forward voltage tends to saturate or exhibit non-monotonic fluctuations. Simply relying on the voltage offset is insufficient to accurately reflect the quality level; therefore, the model automatically reduces the influence of this term. In the luminous flux offset correction term, This is the measured luminous flux. These are standard luminous flux thresholds, all in lumens. This is the relative offset rate of luminous flux. This term is multiplied by a correction factor. , that is to as bottom The natural logarithm of . When hour, The luminous flux offset term has no effect at this time. Determined solely by the voltage term, this corresponds to the general operating condition where, without process damage, the chip's optoelectronic characteristics are primarily graded by voltage. As the degree of damage increases, Increase As the value increases, its logarithm also increases, but the rate of increase slows down. Therefore, the weight of the luminous flux offset term is gradually activated and amplified. This setting reflects the actual performance degradation mechanism of LED devices: the accumulation of process damage leads to the deterioration of the phosphor layer and the decrease in chip light extraction efficiency, making the luminous flux parameter more sensitive to the cumulative effect of damage, and exhibiting a non-linear growth trend as the damage intensifies. The mathematical properties of the logarithmic function can precisely simulate this response curve that is fast at first and then slows down. The sum of the two correction terms yields... This is the comprehensive offset determination coefficient, also dimensionless. Its value and sign comprehensively reflect the degree of photoelectric offset after dynamic correction. This model transforms the originally fixed grading boundary into a boundary determined by the process damage coefficient. The dynamically adjusted judgment baseline ensures that the BIN grading results can truly reflect the quality level of each LED device under the current process conditions.

[0047] The dynamic hierarchical anomaly determination module receives the coupling damage coefficient and the measured photoelectric value, and calculates the BIN offset determination coefficient based on the above model. Then, it is matched against the built-in BIN grading standard table. This standard table is divided into three independent numerical ranges, which are based on a large amount of historical production data. The values ​​are statistically distributed and defined in conjunction with the boundary rules between qualified and defective products. The three intervals correspond to warning-level anomalies, rework-level anomalies, and shutdown-level anomalies, respectively. Products falling into the warning-level anomaly interval... A small absolute value indicates a slight photoelectric offset and that the process damage is still in its early stages; products falling into the rework-level anomaly range... If the value exceeds a certain limit, the photoelectric parameters have deviated significantly, but the device has not completely failed and can be repaired through rework; products falling into the shutdown-level abnormal range, An excessively large value indicates that the process damage has become so severe that it has caused the device to malfunction or poses a reliability risk, necessitating an immediate halt to production. The standard table is a fixed data table structure that is loaded into memory during system initialization. During judgment, a binary search or direct mapping is used to quickly match the relevant range.

[0048] The full-process linkage and traceability module internally constructs three independent control logics corresponding one-to-one with the three levels of anomalies. When a warning-level anomaly signal is received, the first-level control logic is executed, automatically generating a process parameter fine-tuning instruction. The instruction includes the correction amount for the current process temperature setpoint or equipment power output value, and is sent to the equipment controller via the communication interface, ensuring uninterrupted production. When a rework-level anomaly signal is received, the second-level control logic is executed, sending a product batch lock instruction and a rework task order to the manufacturing execution system. The automated diversion device intercepts the batch of products and transfers them to the rework buffer, preventing them from flowing into the next process. When a shutdown-level anomaly signal is received, the third-level control logic is executed, immediately sending a shutdown instruction to the production line main controller, triggering an audible and visual alarm, and displaying process rectification guidelines on the operating terminal. The system can only be restarted after maintenance personnel confirm that the troubleshooting has been completed and the lock has been released. The matching of each level of control logic to the degree of anomaly ensures that the control intensity increases with the gradient of process deterioration, balancing production continuity and quality assurance.

[0049] The full-process coordinated handling and traceability module also has a built-in fixed traceability classification logic. Upon triggering an anomaly, this logic reads the multi-dimensional time-series data collected throughout the entire process corresponding to the abnormal product and the calculated coupling damage coefficient, then executes a classification algorithm. The classification algorithm determines the classification based on the magnitude of each component in the coupling damage coefficient: if... If the value of one item is greater than the other two and exceeds the set dominant threshold, it is determined to be a heat accumulation overload type anomaly, with the source pointing to the process temperature control or heat dissipation link; if The main cause was determined to be a process structure deviation anomaly, with the source pointing to the mechanical positioning system of the die bonding or wire bonding; if The anomaly, identified as a photoelectric parameter drift anomaly, requires confirmation through actual measured photoelectric parameter values. Its source points to material contact characteristics or a drift in the baseline of photoelectric parameter acquisition. The classification result and anomaly level information are combined, along with relevant key acquisition parameters and intermediate calculation values, to generate a standardized electronic traceability report. This report is stored in a structured format in the system database, allowing process engineers to query and analyze it through a report interface, providing data support for subsequent process parameter optimization and preventative maintenance.

[0050] This application also provides a method for quality control of the entire LED process. This method runs on the aforementioned system and is executed sequentially by an industrial measurement and control processor. Step S1: Receive the input target LED product model parameters, search the parameter adaptation database and retrieve the matching thermal accumulation influence weight coefficient, structural offset weight coefficient, contact length weight coefficient, maximum allowable thermal accumulation threshold, and maximum process offset threshold. Write these parameters in batches into the configuration registers of the thermal accumulation-structure coupling calculation module and the dynamic hierarchical anomaly judgment module to complete the initialization of the system control model. Step S2: During the execution of each process of LED die bonding, wire bonding, packaging, and color sorting, the multi-dimensional time-series acquisition module of the entire process synchronously acquires the temperature data sequence, equipment power loss data sequence, die bonding offset, wire bonding offset, effective contact length, measured forward voltage, and measured luminous flux of each process. Bind all data to a unique product control ID and organize them into a multi-dimensional time-series dataset. Step S3: The thermal accumulation-structure coupling calculation module extracts thermal parameters from the multi-dimensional time-series dataset, calls the time-series thermal integration calculation logic, runs the multi-process time-series superimposed thermal accumulation coefficient calculation model, performs integration calculations on the time-series data of each process, solves for the corresponding thermal accumulation impact, and determines whether the thermal damage state of the process exceeds the allowable range based on the maximum allowable thermal accumulation threshold. Step S4: Integrating the thermal accumulation impact, process structure offset, and effective contact length data obtained in the previous step, the module calls the multi-parameter normalization coupling calculation logic, runs the multi-dimensional process coupling structure damage coefficient calculation model, and outputs the coupling damage coefficient through normalization and weighted summation operations. Step S5: The dynamic grading anomaly judgment module receives the coupling damage coefficient, obtains the measured values ​​of photoelectric parameters, calls the photoelectric parameter correction input logic, runs the BIN photoelectric parameter dynamic judgment model based on process damage coupling, calculates the BIN offset judgment coefficient, and then matches this coefficient with the three sets of numerical intervals in the BIN grading judgment standard table to lock the production anomaly level. Step S6: The full-process linkage handling and traceability module selects the corresponding control logic based on the anomaly level, and executes fine-tuning of process parameters, product batch locking, and rework or production line shutdown rectification operations. Step S7: The abnormal batch of products after handling is retested, the photoelectric parameters after retesting are obtained and the process status is verified. At the same time, the source tracing classification logic is activated to locate the type of anomaly source, generate an electronic source tracing report, and iteratively update the process thresholds and weight parameters in the system based on the analysis results, thereby optimizing the system control model.

[0051] In the entire computational process of this method, steps S3, S4, and S5 form a progressively advancing closed-loop computational chain. The cumulative thermal impact output by step S3 is a necessary input for the multi-parameter normalization coupling calculation in step S4. The coupling damage coefficient output by step S4 is the sole basis for correcting the BIN offset determination coefficient in step S5. The determination result of step S5 is then fed back to the process threshold iteration in step S1 through control and traceability feedback, forming a continuous optimization closed loop. This progressive relationship ensures that all determination results can be traced back to the original instantaneous temperature and power consumption acquisition values, guaranteeing the full-chain traceability of quality control data and avoiding parameter fragmentation between independent computational stages.

[0052] The implementation method of this application will be further described below with reference to a specific production scenario. A certain LED packaging production line plans to produce a product with the model number "SMD2835". The operator inputs this model code through a human-machine interface terminal. After receiving the code, the product parameter dynamic configuration and model change module performs a precise matching query in the parameter adaptation database. The database returns the complete parameter set corresponding to this model, including... , , , , , , , And the standard voltage of this model in the spectral separation and color sorting process. Standard luminous flux These parameters are synchronously loaded into the computational variable areas of the thermal accumulation-structure coupling calculation module and the dynamic hierarchical anomaly determination module via memory mapping, thus completing the initialization of the system control model.

[0053] When the die bonding process starts, the multi-dimensional timing acquisition module generates a batch control ID for the batch of products to be produced. The ID's structure is "date code + production line number + serial number," and it is pre-assigned to a virtual ID for each individual LED within that batch. As production begins, the timing thermal data acquisition submodule continuously collects temperature data from the die bonding machine's heating platform near the chip placement area at a sampling period of 0.2 seconds. Simultaneously, it collects the active power loss of the heating unit and ultrasonic transducer via a smart meter. The process structure data acquisition submodule captures an image the instant after chip mounting is completed by a high-precision industrial camera. Using an edge detection algorithm, it locates the chip outline and alignment marks on the substrate, calculating the Euclidean distance between the chip center and the mark center as the offset. Simultaneously, the effective contact length is obtained by measuring the area of ​​the solder spreading region and converting it into the equivalent circle diameter. The first LED in this batch had 150 temperature points and 150 power points collected during the die bonding process, and the actual measurements were... , All data is tagged with the same control ID and packaged into data frames for transmission.

[0054] Upon receiving the data frame, the thermal accumulation-structure coupling calculation module first initiates the numerical integration program via the time-series thermal integration calculation logic. This program iterates through the time-series data, performing numerical integration for each sampling point. ,calculate Then multiply by the sampling interval of 0.2 seconds, and sum them up. Where... The standard temperature for the die bonding process is 180 degrees Celsius, converted to Kelvin temperature difference. The cumulative heat effect of the die bonding process for this LED is calculated. Subsequently, the multi-parameter normalization coupled computation logic is called. , , And the weights, perform the calculation: The coupling damage coefficient is transmitted to the dynamic hierarchical anomaly determination module.

[0055] The calculation process for subsequent wire bonding steps is similar; the LED was measured during the wire bonding process. , , The calculation yields The system follows a progressive process principle, taking the maximum value among all processes. Value or multiple processes The values ​​are then weighted and synthesized. Here, the principle of maximum damage is taken as the comprehensive value, and 0.701 of the die bonding process is taken as the final coupling damage coefficient of the product.

[0056] After the LED enters the spectral sorting and color separation process, the photoelectric parameter acquisition submodule measures the forward voltage. luminous flux The dynamic hierarchical anomaly detection module will... , , , , Substitute into the dynamic determination model of BIN photoelectric parameters: The coefficient falls within the first value range of the standard table, corresponding to a warning-level anomaly. The system then triggers the first-level control logic, generating a voltage adjustment fine-tuning command to prompt the operator to pay attention to the reference voltage setting of the spectrophotometer, and production resumes normal operation.

[0057] In another example, the production line switched to produce "5050" model LEDs, whose parameter set is... , , , , , , , , , The cumulative thermal effect of an LED during the die bonding process was measured. It has exceeded the threshold. structural offset , ,calculate Measured by spectrophotometry , Substitute If this coefficient falls within the rework-level anomaly range, the entire process linkage and traceability module will immediately lock onto the batch and issue a rework order. (Source tracing and classification logic analysis) The contribution of each component was significant, with the heat accumulation component contributing 0.525, which was the dominant factor. This was determined to be a heat accumulation overload anomaly, with the source pointing to temperature parameter drift or heat dissipation abnormality in the die bonding process. The generated traceability report detailed all the data related to this anomaly, guiding technicians to inspect the temperature control system of the die bonding machine.

[0058] As can be seen from the above description, the system and method provided in this application incorporate the thermal, structural and photoelectric parameters of the entire LED production process into a unified and coupled numerical calculation framework. By using integration, normalization and dynamic correction rules, the traditional fixed threshold quality judgment is transformed into an adaptive judgment based on the cumulative effect of process damage. Furthermore, a differentiated control and automatic traceability mechanism matching the anomaly level is established, so that the quality control of the entire LED process forms a complete closed loop.

[0059] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A full-process intelligent control system for LEDs with spectral and color separation BIN-level linkage, characterized in that, It includes a thermal accumulation-structure coupling calculation module, a dynamic hierarchical anomaly judgment module, and a full-process linkage handling and traceability module; The thermal accumulation-structure coupling calculation module is used to receive multi-dimensional time-series acquisition data corresponding to the entire process of LED die bonding, wire bonding, packaging, and color sorting; calculate the thermal accumulation impact based on the thermal-related parameters inside the multi-dimensional time-series acquisition data; and calculate the coupling damage coefficient based on the thermal accumulation impact and the structural parameters inside the multi-dimensional time-series acquisition data. The dynamic hierarchical anomaly determination module is communicatively connected to the thermal accumulation-structure coupling calculation module. It is used to receive the coupling damage coefficient, calculate the BIN offset determination coefficient based on the coupling damage coefficient and the measured value of LED photoelectric parameters, and lock the production anomaly level according to the numerical range corresponding to the BIN offset determination coefficient. The full-process linkage handling and traceability module is communicatively connected to the dynamic hierarchical anomaly judgment module. It is used to receive the production anomaly level, trigger production control operations that match the production anomaly level, and locate the process source type that caused the production anomaly.

2. The LED full-process intelligent control system with spectral and color separation BIN-level linkage according to claim 1, characterized in that, It also includes a product parameter dynamic configuration and switching module, which has a built-in parameter adaptation database; The product parameter dynamic configuration and model change module is used to receive the target LED product model parameters entered externally, retrieve the thermal accumulation influence weight coefficient, structural offset weight coefficient, contact length weight coefficient, maximum allowable thermal accumulation threshold, and maximum process offset threshold that match the target LED product model parameters in the parameter matching database, and synchronously transmit all matching parameters to the thermal accumulation-structure coupling calculation module and the dynamic hierarchical anomaly judgment module.

3. The LED full-process intelligent control system with spectral and color separation BIN-level linkage according to claim 1, characterized in that, It also includes a full-process multi-dimensional time-series acquisition module, which is communicatively connected to the thermal accumulation-structure coupling calculation module and the product parameter dynamic configuration and change module. The full-process multi-dimensional time-series acquisition module includes a time-series thermal data acquisition sub-module, a process structure data acquisition sub-module, and a photoelectric parameter acquisition sub-module. The time-series thermal data acquisition submodule is used to collect process temperature data and equipment power loss data throughout the entire LED production process. The process structure data acquisition submodule is used to acquire the offset amount of the LED die bonding process, the offset amount of the wire bonding process, and the effective contact length of the LED manufacturing process. The photoelectric parameter acquisition submodule is used to acquire the measured values ​​of the forward voltage and luminous flux of the LED device. The full-process multi-dimensional time-series acquisition module binds a unique product management ID to all acquired data, generates full-process multi-dimensional time-series acquisition data, and completes data output.

4. The LED full-process intelligent control system with spectral and color separation BIN-level linkage according to claim 1, characterized in that, The thermal accumulation-structure coupling calculation module is configured with progressive linkage calculation logic, which includes time-series thermal integral calculation logic, multi-parameter normalization coupling calculation logic, and photoelectric parameter correction input logic. The time-series heat integral calculation logic is used to perform time integral calculations on the time-series heat-related parameters of the entire process and output the corresponding cumulative heat influence. The multi-parameter normalized coupling calculation logic is used to receive the thermal cumulative influence, the process structure offset, and the effective contact length of the process, to complete the multi-dimensional process parameter normalized coupling calculation, and to output the corresponding coupling damage coefficient. The photoelectric parameter correction input logic is used to transmit the coupling damage coefficient to the dynamic hierarchical anomaly determination module in a directional manner, providing a fixed input variable for the photoelectric parameter offset correction calculation.

5. The LED full-process intelligent control system with spectral and color separation BIN-level linkage according to claim 4, characterized in that, The time-series thermal integration calculation logic configures the process time-series thermal accumulation quantization operation rules, the multi-parameter normalization coupling calculation logic configures the process damage quantization operation rules, and the photoelectric parameter correction input logic configures the BIN level offset quantization operation rules. The calculation dimensions of the process timing thermal accumulation quantification calculation rules, process damage quantification calculation rules, and BIN level offset quantification calculation rules are mutually compatible. All calculation rules rely on the LED full-process production parameters to establish a normalized linkage calculation system, realizing the linkage quantification processing of thermal parameters, structural parameters, and photoelectric parameters.

6. The LED full-process intelligent control system for beam splitting and color separation BIN-level linkage according to claim 1, characterized in that, The dynamic hierarchical anomaly determination module has a built-in fixed BIN hierarchical determination standard table. The BIN hierarchical determination standard table is divided into three independent numerical intervals. The three independent numerical intervals correspond one-to-one with the numerical values ​​of the BIN offset determination coefficient. The three independent numerical intervals are respectively matched with three anomaly levels: warning level anomaly, rework level anomaly, and shutdown level anomaly.

7. The LED full-process intelligent control system for beam splitting and color separation BIN-level linkage according to claim 6, characterized in that, The full-process linkage and traceability module has a built-in three-level independent control logic, which corresponds one-to-one with the early warning level anomaly, the rework level anomaly, and the shutdown level anomaly. The three independent control logics are used to push out instructions for fine-tuning process parameters, locking and rework of product batches, and shutdown and process rectification of production lines, respectively, to complete control operations such as correcting minor process deviations, intercepting non-conforming products, and shutting down and rectifying production line malfunctions.

8. The LED full-process intelligent control system with spectral and color separation BIN-level linkage according to claim 1, characterized in that, The full-process linkage handling and traceability module has a built-in fixed source classification logic. The source classification logic reads multi-dimensional time-series data collected from the entire process and the coupling damage coefficient, and matches and determines three types of process abnormality sources: thermal accumulation overload type, process structure deviation type, and photoelectric parameter drift type. The full-process linkage and traceability module integrates anomaly level data and process source type data to generate a standardized electronic traceability report and completes storage and archiving.

9. A method for intelligent control of the entire process of LEDs with spectral and color separation BIN-level linkage, characterized in that, The intelligent control system for the entire LED process, applied to the BIN-level linkage system for spectral and color separation as described in any one of claims 1 to 8, is executed by an industrial measurement and control processor and includes the following steps: S1. Enter the model parameters of the target LED product, retrieve the matching weight coefficients and process threshold parameters, and complete the adaptation configuration of the system control model. S2. Collect thermal, structural, and optoelectronic parameters of the entire LED manufacturing process, including die bonding, wire bonding, packaging, and color sorting, and construct a multi-dimensional time-series dataset bound to the product management ID; S3. Solve the cumulative thermal impact by integrating the thermal parameters of the multidimensional time series dataset, and determine the thermal damage status of the process by combining the preset threshold. S4. Integrate the cumulative thermal impact, process structure offset data, and effective contact length data to calculate the coupling damage coefficient; S5. Based on the coupling damage coefficient, the photoelectric parameter judgment standard is corrected, the BIN offset judgment coefficient is calculated, and the abnormality level is determined by matching the numerical range. S6. Perform corresponding process fine-tuning, batch rework, and production line shutdown rectification operations according to the abnormality level; S7. Retest the parameters of abnormal batches of products and verify the process status, locate the source of process abnormality, and iteratively update the system process thresholds and weight parameters.

10. The method for intelligent control of the entire LED process with BIN-level linkage based on spectral and color separation as described in claim 9, characterized in that, Steps S3, S4, and S5 form a progressive closed-loop calculation process. The quantitative data output from the preceding steps is the only core input for the subsequent steps. The entire process realizes the linkage closed-loop determination of thermal parameters, structural parameters, and photoelectric parameters of the entire LED process, and completes the synchronous execution of process damage quantification and BIN classification dynamic correction.