Semiconductor packaging test optimization method and system
By collecting and analyzing multi-dimensional parameters of packaging and testing equipment in real time and dynamically adjusting the load, the inefficiency of equipment management in traditional packaging and testing is solved, achieving precise control of equipment status and load optimization, thereby improving production efficiency and equipment lifespan.
Patent Information
- Application Number
- CN202511905732.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-12-17
- Publication Date
- 2026-03-06
AI Technical Summary
In traditional semiconductor packaging testing, equipment operation management and load distribution are driven by experience or controlled by fixed parameters, which makes it difficult to meet the needs of high-efficiency and high-precision production, resulting in low equipment utilization, decreased testing efficiency and increased failure rate.
By collecting multi-dimensional operating parameters in real time, using the analytic hierarchy process (AHP) and eigenvalue method to calculate equipment state preference values, dynamically adjusting the semiconductor feeding amount, and combining historical data to divide state intervals and construct performance degradation curves, precise equipment management and load optimization can be achieved.
It improves equipment stability, reduces failure risk, increases production line utilization, extends equipment lifespan, reduces maintenance costs, and improves packaging and testing yield.
Smart Images
Figure CN121615369A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of semiconductor manufacturing technology, and in particular to an optimized method for semiconductor packaging and testing. Background Technology
[0002] Against the backdrop of the rapid development of the semiconductor industry, packaging and testing, as a key back-end process in semiconductor manufacturing, directly affects chip yield, reliability, and production efficiency. With the continuous improvement of chip integration (such as advanced process chips having more than a thousand pins) and the evolution of packaging forms towards miniaturization (such as CSP, WLCSP) and heterogeneous integration (such as SiP), higher requirements are placed on the operating accuracy, stability, and load adaptability of packaging and testing equipment. Currently, in semiconductor packaging and testing production, equipment operation management and load distribution are still mainly driven by traditional experience or fixed parameter control, which has the following core technical problems and makes it difficult to meet the needs of high-efficiency and high-precision production.
[0003] In traditional semiconductor packaging testing, a fixed number of semiconductors are typically allocated to each packaging and testing device. This method does not take into account the differences in the real-time operating status of each device. When some devices are in good working condition, insufficient semiconductors may lead to low device utilization; while when some devices are in poor working condition, excessive semiconductors may lead to decreased testing efficiency and increased failure rate. To address this, we propose an optimization method for semiconductor packaging testing. Summary of the Invention
[0004] To address the shortcomings of the existing technologies, the technical problem to be solved by this invention is to propose a semiconductor packaging testing optimization method that can fit a decay curve by analyzing the changes in core parameters (such as vibration amplitude) of equipment over the past 3 months. This method can intuitively present the performance decay rhythm over the next 1-2 months and accurately calculate the estimated time for core parameters to drop to the threshold of the next level of state. This allows maintenance to shift from post-fault repair to pre-fault prediction. For example, if it is known in advance that the vibration amplitude of a certain piece of equipment will reach the warning threshold in 1 month, maintenance can be arranged in advance to avoid production interruptions caused by sudden failures.
[0005] One technical solution adopted in this invention is: a semiconductor packaging testing optimization method, which includes the following steps: S1: Real-time acquisition of operating parameters of the current packaging and testing equipment based on sensors; S2: After preprocessing and analyzing the acquired operating parameters, determine the weight values of different indicators, and calculate the current state preference value of the packaging and testing equipment based on the weight values of the indicators. S3: Based on historical packaging and testing equipment data, construct a range of state preference values, and substitute the obtained state preference values into different ranges; S4: Dynamically adjust the semiconductor loading amount of the packaging and testing equipment according to the state preference value range in which the packaging and testing equipment is located.
[0006] Furthermore, the sensors include a laser displacement sensor, a torque sensor, a vibration sensor, a temperature sensor, and a current sensor; the laser displacement sensor is deployed at the wafer stage and packaging mold positioning point to detect positioning and offset parameters; the torque sensor is deployed at the output shaft of the drive motor to detect speed and rotational speed parameters; the vibration sensor is deployed at the equipment frame, motor housing, and guide rail base to detect force and vibration parameters; the temperature sensor is deployed in the welding area to detect temperature parameters; and the current sensor is deployed at the main power supply circuit of the equipment to detect power consumption parameters.
[0007] Furthermore, the analysis of the acquired operating parameters after preprocessing includes the following sub-steps: S21: Perform data cleaning, outlier removal, and data standardization on the operating parameters; S22: Based on the preprocessed operating parameters, construct a three-level evaluation index: target layer, criterion layer, and indicator layer. The target layer is the comprehensive equipment status evaluation, the criterion layer is the mechanical system, environmental process, electrical performance and special process dimensions, and the indicator layer is the specific parameters under each criterion layer.
[0008] Furthermore, determining the weight values of different indicators includes the following sub-steps: S23: The weight values are determined by the analytic hierarchy process (AHP) and a judgment matrix is constructed. Experts in the field of semiconductor packaging and testing are invited to score the importance of the criteria layer and the indicator layer based on the 1-9 scale to form a judgment matrix. S24: Calculate the consistency ratio CR of the judgment matrix. If CR < 0.1, the test is passed. If the test is not passed, feedback is given to the experts to adjust the judgment matrix until the test is passed. S25: Using the eigenvalue method, calculate the eigenvector corresponding to the largest eigenvalue of the judgment matrix to obtain the subjective weights of the criterion layer and the subjective weights of the indicator layer. S26: Calculate objective weights based on historical evaluation data: ; ; in, This represents the percentage of the standardized value of the i-th indicator in the j-th period, where n represents the number of periods. Let represent the entropy value of the i-th evaluation index, and ln denote the natural logarithm operation. This represents the objective weight of the i-th indicator; S27: Combine the subjective weights and objective weights in a 7:3 ratio to obtain the final weights: ; in, This represents the final weight of the i-th indicator. This represents the subjective weight of the i-th indicator.
[0009] Furthermore, the specific calculation process for calculating the current state preference value of the packaging and testing equipment based on the weight values of the indicators is as follows: ; in, This represents the state preference value of the k-th criterion layer. This represents the number of indicators in the k-th criterion layer. This represents the standardized value of the i-th indicator in the k-th criterion layer. This represents the final weight of the i-th indicator.
[0010] Furthermore, step S3 includes the following sub-steps: S31: Based on the different device states in the historical packaging and testing equipment data, divide the intervals according to percentiles, taking P90, P60 and P40 as boundaries. The historical packaging and testing equipment data includes packaging and testing equipment data in normal working state, warning state and fault state. S32: Substitute the calculated state preference value into different intervals to classify and judge the state of the current packaging and testing equipment.
[0011] Furthermore, step S4 includes the following sub-steps: S41: Generate corresponding control commands based on the location of different packaging and testing equipment, and send the control commands to the corresponding packaging and testing equipment to adjust the semiconductor feeding amount; S42: Repeat steps S1-S4 within a preset interval to readjust the semiconductor loading amount of the packaging and testing equipment.
[0012] Furthermore, before step S41, the following sub-steps are also included: S40: Pre-establish a mapping library between the state range of the packaging and testing equipment and the semiconductor loading parameters.
[0013] Furthermore, step S4 also includes the following sub-steps: S43: Construct a performance degradation curve model based on the operating data of the packaging and testing equipment throughout its entire lifecycle; S44: Calculate the attenuation rate of each package test device under the current load, and allocate the load to each package test device based on the attenuation rate and the equalization factor.
[0014] The second technical solution adopted in this invention is: a semiconductor packaging and testing optimization system, which includes the following modules: The parameter acquisition module is used to acquire the operating parameters of the current packaging and testing equipment in real time based on sensors. The preference value calculation module is used to preprocess and analyze the acquired operating parameters, determine the weight values of different indicators, and calculate the current state preference value of the packaging and testing equipment based on the weight values of the indicators. The device partitioning module is used to construct a range of state preference values based on historical packaging and testing equipment data, and then substitute the obtained state preference values into different ranges. The equipment adjustment module is used to dynamically adjust the semiconductor loading amount of the packaging and testing equipment according to the state preference value range in which the packaging and testing equipment is located.
[0015] The semiconductor packaging testing optimization method of the present invention has at least the following beneficial effects: It collects multi-dimensional operating parameters in real time using various sensors, promptly captures anomalies, eliminates monitoring blind spots, reduces equipment failure risks, and ensures stable operation; it avoids subjective misjudgments and provides accurate basis for decision-making by using weighted calculations and quantified state preference values that integrate subjective and objective data; it scientifically divides three-level state intervals based on historical data to achieve hierarchical equipment control and improve management efficiency; it dynamically adjusts the number of semiconductors on a single device to match equipment performance and release production capacity; and it prioritizes the allocation of devices with low attenuation rates among multiple devices in similar states according to a load balancing coefficient, improving the utilization rate of the entire production line. In the long term, it can also reduce equipment wear, extend the life cycle, reduce procurement and maintenance costs, and improve packaging testing yield through precise state control, thus achieving efficiency, cost, and quality benefits. Attached Figure Description
[0016] The accompanying drawings, which are included to provide a further understanding of this application and form part of this application, illustrate exemplary embodiments and are used to explain this application, but do not constitute an undue limitation of this application. In the drawings: Figure 1 This is a flowchart of a semiconductor packaging testing optimization method according to the present invention.
[0017] Figure 2 for Figure 1 The sub-flowchart of step S2.
[0018] Figure 3 for Figure 1 The sub-flowchart of step S3 in the middle.
[0019] Figure 4 for Figure 1 The sub-flowchart of step S4 in the middle.
[0020] Figure 5This is a block diagram of a semiconductor packaging testing optimization system according to the present invention. Detailed Implementation
[0021] The invention will now be further described with reference to the accompanying drawings.
[0022] Please see Figure 1 The above is a flowchart of a semiconductor packaging testing optimization method according to the present invention. The method may include the following steps: S1: Real-time acquisition of operating parameters of the current packaging and testing equipment based on sensors.
[0023] In step S1, the sensors include a laser displacement sensor, a torque sensor, a vibration sensor, a temperature sensor, and a current sensor. The laser displacement sensor is deployed at the wafer stage and packaging mold positioning point, the torque sensor is deployed at the drive motor output shaft, the vibration sensor is deployed at the equipment frame, motor housing, and guide rail base, the temperature sensor is deployed in the welding area, and the current sensor is deployed at the equipment's main power supply circuit. The detected operating parameters include: positioning and offset parameters, speed and rotational speed parameters, force and vibration parameters, temperature parameters, and power consumption parameters.
[0024] The laser displacement sensor in this application is precisely deployed at the wafer stage and packaging mold positioning points, which can directly capture positioning and offset parameters, avoiding insufficient packaging accuracy caused by positioning deviation; the torque sensor is mounted on the output shaft of the drive motor, which can acquire speed and rotation parameters in real time, accurately reflecting the motor's operating status; the vibration sensor is arranged at multiple points on the rack, motor housing, and guide rail base, which can comprehensively collect force and vibration parameters, avoiding the limitations of single-point monitoring; the temperature sensor focuses on the welding area, which can timely grasp the temperature of key processes; and the current sensor monitors the total power supply circuit, ensuring accurate acquisition of energy consumption parameters and providing a reliable data foundation for subsequent analysis. In terms of the comprehensiveness of equipment status coverage, the solution covers five core dimensions: positioning, power, structure, process, and energy consumption. By collecting key parameters through corresponding sensors, it can comprehensively grasp the status of the equipment from the operation of core components to the overall energy consumption, avoiding the omission of potential equipment hazards due to the lack of monitoring dimensions, and providing comprehensive data support for subsequent status preference value calculation, ensuring a more objective and accurate judgment of the equipment's working status.
[0025] S2: After preprocessing the acquired operating parameters, analyze them to determine the weight values of different indicators, and calculate the current state preference value of the packaging and testing equipment based on the weight values of the indicators. Preprocessing in step S2 includes data cleaning, outlier removal, and data standardization. Based on the preprocessed parameters, construct a three-level evaluation index system: target layer, criterion layer, and indicator layer. The target layer is the comprehensive equipment state evaluation; the criterion layer includes mechanical system, environmental process, electrical performance, and special process dimensions; and the indicator layer consists of the specific parameters under each criterion layer.
[0026] Please see Figure 2 This is a sub-flowchart for step S2. The "analysis after preprocessing the acquired operating parameters" in step S2 may include the following sub-steps: S21: Perform data cleaning, outlier removal, and data standardization on the operating parameters. The data preprocessing in step S21 ensures the accuracy of the analysis. Data cleaning removes redundant data caused by signal interference during sensor acquisition. Outlier removal avoids biased data caused by momentary equipment failures or sensor mis-touch. Data standardization unifies the units and numerical ranges of different parameters, avoiding the impact of parameter unit differences on weight calculation and comprehensive analysis. The combination of these three steps ensures the purity and consistency of the input data from the source, laying a reliable data foundation for subsequent evaluation index calculations and reducing the interference of invalid data on status judgment.
[0027] S22: Based on the preprocessed operating parameters, a three-tiered evaluation index system is constructed: a target layer, a criterion layer, and an indicator layer. The target layer assesses the overall equipment condition; the criterion layer covers mechanical systems, environmental processes, electrical performance, and special processes; and the indicator layer comprises the specific parameters under each criterion layer. Step S22 achieves a systematic evaluation through this three-tiered index system. The target layer clarifies the core direction of the overall equipment condition assessment, preventing the assessment from deviating from core requirements. The criterion layer divides dimensions according to mechanical systems, environmental processes, electrical performance, and special processes, accurately covering key areas of equipment operation and ensuring no dimension is missed in the assessment. The indicator layer breaks down the criterion layer into specific parameters, achieving a layered implementation from macro-level objectives to meso-level dimensions to micro-level parameters, making the evaluation logic clear and traceable, and preventing the overall condition analysis from becoming vague.
[0028] The "determining the weight values of different indicators" step in S2 may include the following sub-steps: S23: The weight values are determined using the analytic hierarchy process (AHP), a judgment matrix is constructed, and experts in the field of semiconductor packaging and testing are invited to score the importance of the criterion layer and the indicator layer based on the 1-9 scale to form a judgment matrix. In this step S23, the weights of each indicator are allocated, the weight values are determined using the AHP, a judgment matrix is constructed, and experts in the field of semiconductor packaging and testing are invited to score the importance of the criterion layer and the indicator layer based on the 1-9 scale to form a judgment matrix.
[0029] S24: Calculate the consistency ratio CR of the judgment matrix. If CR < 0.1, the test is passed. If the test is not passed, feedback is given to the experts to adjust the judgment matrix until the test is passed. S25: Using the eigenvalue method, calculate the eigenvector corresponding to the largest eigenvalue of the judgment matrix to obtain the subjective weights of the criterion layer and the subjective weights of the indicator layer. S26: Calculate objective weights based on historical evaluation data: ; ; in, This represents the percentage of the standardized value of the i-th indicator in the j-th period, where n represents the number of periods. Let represent the entropy value of the i-th evaluation index, and ln denote the natural logarithm operation. This represents the objective weight of the i-th indicator; S27: Combine the subjective weights and objective weights in a 7:3 ratio to obtain the final weights: ; in, This represents the final weight of the i-th indicator. This represents the subjective weight of the i-th indicator.
[0030] The specific calculation process for "calculating the current state preference value of the packaging and testing equipment based on the weight values of the indicators" in step S2 is as follows: ; in, This represents the state preference value of the k-th criterion layer. This represents the number of indicators in the k-th criterion layer. This represents the standardized value of the i-th indicator in the k-th criterion layer. This represents the final weight of the i-th indicator.
[0031] The steps S23-S27 above describe the specific process of weight value calculation. This scheme has a rigorous weight determination logic, taking into account both subjective experience and objective data. It constructs a judgment matrix by combining the analytic hierarchy process with the expert 1-9 scaling method, incorporating domain professional experience to ensure that the weights meet the actual needs of the packaging and testing scenario. The consistency check (CR < 0.1) avoids logical contradictions in expert scoring and ensures the rationality of subjective weights. At the same time, it calculates the entropy value based on historical data to obtain objective weights, reflecting the information contribution of the indicators in actual operation. The 7:3 subjective-objective integration ratio retains the expert experience's control over core indicators and corrects subjective biases through data patterns, making the weight values more in line with the actual operating characteristics of the equipment. The comprehensive state preference value calculation accurately quantifies the equipment state. The weighted sum logic of standardized value × final weight transforms the scattered multi-dimensional parameters (such as temperature, vibration, and energy consumption) into comparable quantitative indicators, realizing the transformation of equipment state from fragmented description of multiple parameters to single-value comprehensive evaluation.
[0032] S3: Based on historical packaging and testing equipment data, construct state preference value intervals and substitute the obtained state preference values into different intervals. In step S3, different state preference value intervals are established by using historical packaging and testing equipment data, including data from normal operation, warning, and fault states. Intervals are divided according to percentiles, with P90, P60, and P40 as boundaries, corresponding to intervals of Excellent ≥90, Warning 60-90, and Fault <40. The calculated state preference values are then substituted into different intervals to classify and judge the current state of the packaging and testing equipment.
[0033] Please see Figure 3 This is a sub-flowchart of step S3 of the present invention. Step S3 may include the following sub-steps: S31: Based on the different device states in historical packaging and testing equipment data, intervals are divided according to percentiles, using P90, P60, and P40 as boundaries. The historical packaging and testing equipment data includes data from normal operation, warning, and fault states. This S31 step constructs intervals based on the three core states of normal, warning, and fault states in historical data, rather than relying on theoretical assumptions. This ensures that the intervals accurately reflect the actual operating states of the equipment. The fault interval corresponds to historical fault data characteristics, and the warning interval matches past sub-health states requiring intervention. This makes state judgment more practically instructive, avoiding the risk of misjudging normal or missing judgments. Using P90, P60, and P40 as interval boundaries, and determining thresholds through data distribution patterns rather than subjectively setting values, effectively avoids human experience bias. Furthermore, percentiles have strong inclusiveness for abnormal data and can stably reflect the state distribution characteristics of most devices, providing a unified and comparable evaluation standard for the state preference values of different devices, avoiding judgment confusion caused by ambiguous boundaries.
[0034] S32: Substitute the calculated state preference value into different intervals to classify and determine the current state of the packaging and testing equipment. This S32 step is efficient and clearly directional in state determination. By directly substituting the calculated state preference value into the interval, the current level of the equipment (Excellent-Warning-Fault) can be quickly located without complex secondary analysis. At the same time, different intervals correspond to clear subsequent processing directions, providing accurate decision-making basis for the dynamic adjustment of S4 and improving equipment management efficiency.
[0035] S4: Dynamically adjust the semiconductor loading amount of the packaging and testing equipment according to the state preference value range in which the packaging and testing equipment is located.
[0036] Please see Figure 4 This is a sub-flowchart of step S4 of the present invention. Step S4 may include the following sub-steps: S40: Pre-establish a mapping library between the state intervals of the packaging and testing equipment and the semiconductor loading parameters. The specific logic of this mapping relationship is as follows: The instruction parameters include the basic material quantity, the standard load corresponding to the equipment status range, the feeding rhythm parameters, the feeding interval time, and the quantity pushed by the feeder in a single push. Verify the parameters. After loading, the equipment waits for a confirmation signal to verify the execution effect of the instructions. The system generates an adjustment command only when the device status meets the following conditions: The current state range of the device has changed compared to the range corresponding to the last instruction; The state range remains unchanged, but real-time parameter fluctuations exceed the threshold, triggering a compensatory adjustment; Command priority rules: When both range change adjustment and parameter fluctuation adjustment are triggered simultaneously on the same device, commands are generated in order of urgency.
[0037] The pre-established mapping database clearly defines the core parameters of basic material quantity and feeding rhythm corresponding to different state intervals. This eliminates the need for temporary calculations or reliance on manual decision-making when generating instructions, allowing the equipment management system to quickly match intervals and generate standardized instructions. This avoids parameter configuration deviations, significantly shortens instruction generation time, and improves decision-making efficiency. The execution effect of instructions is verifiable, ensuring the reliability of adjustments. After feeding, parameters are verified by waiting for equipment feedback confirmation signals, which can promptly detect instruction execution anomalies and avoid blind adjustments that end as soon as instructions are issued. If the verification fails, instructions can be reissued in a timely manner, ensuring that every adjustment of the feeding quantity is accurately implemented and that the equipment operates as expected.
[0038] In addition, two trigger conditions can be set: range change and parameter fluctuation exceeding the threshold. The command is only generated when the device status really needs to be adjusted. This prevents frequent invalid adjustments when the device status is stable, and can respond in a timely manner when the status changes or parameters are abnormal, thus balancing the necessity and economy of adjustment.
[0039] S41: Generate corresponding control commands based on the location of different packaging and testing equipment, and send the control commands to the corresponding packaging and testing equipment to adjust the semiconductor feeding amount; S42: Repeat steps S1-S4 within a preset interval to readjust the semiconductor loading amount of the packaging and testing equipment. In these S41-S42 steps, the data on the semiconductors on the equipment is dynamically adjusted. The equipment management system generates control commands based on the packaging and testing equipment status for different intervals, which are then sent to the equipment PLC via industrial Ethernet. The equipment can complete parameter adaptation and adjust the semiconductor loading amount within 10 seconds, significantly shortening the "status judgment-adjustment" interval. Whether the equipment is in a warning state requiring a reduction in loading to avoid overload, or in an optimal state requiring an increase in loading to improve efficiency, it can respond quickly, avoiding equipment damage or efficiency waste caused by adjustment lag, and ensuring the equipment is always in optimal operating condition. Furthermore, operating parameters can be re-acquired and analyzed every 30 minutes, and the semiconductor loading amount can be readjusted based on the analysis results. In terms of operational stability, the operating parameters are re-collected and adjusted every 30 minutes to form a closed loop of real-time monitoring, dynamic adaptation and periodic optimization. This high-frequency dynamic calibration can respond to parameter fluctuations during equipment operation in a timely manner, avoid operational deviations caused by parameter drift after a single adjustment, maintain stable operation of the equipment in the high-efficiency range for a long time, reduce test errors or failure risks caused by uncontrolled conditions, and ensure the consistency of packaging and testing quality.
[0040] S43: Based on the operational data of the packaging and testing equipment throughout its entire lifecycle, construct a performance degradation curve model. In this S43 step, an inter-device load balancing coefficient is introduced. By collecting data on load, runtime, and key parameter changes throughout the equipment's lifecycle, the dominant degradation factors are screened and standardized. A basic database is constructed, and a performance degradation curve model is built for each device. The degradation rate under the current load is evaluated. Combined with the real-time parameter correction model, a comprehensive coefficient is calculated and its consistency is verified, with the degradation rate as the core and combined with maintenance records and load adaptability as the balancing coefficient dimensions. Devices are sorted by coefficient from high to low, and the load is preferentially allocated to devices with high coefficients. The incremental load is controlled to not exceed the optimal upper limit. After allocation, the parameters are monitored in real time, and the allocation effect is reviewed periodically. The balancing coefficient dimensions, weights, and allocation rules are optimized, and the load is preferentially allocated to devices with low degradation rates to extend the stable operation cycle of the equipment. By collecting data from the entire lifecycle of the equipment to build a performance degradation curve model, it is possible to accurately assess the degradation rate of each device under the current load. Combined with maintenance records and load adaptability, a comprehensive balancing coefficient is determined. The load is allocated according to the coefficient, which can prioritize the allocation of semiconductors to devices with low degradation rates and high adaptability. This avoids the problem of some devices being overloaded and degraded, while others are idle and wasted due to average allocation. This achieves the optimal matching of load and performance of the equipment cluster and improves the overall operating efficiency.
[0041] S44: Calculate the attenuation rate of each packaging and testing device under the current load, and allocate the load to each packaging and testing device based on the attenuation rate and the equalization factor. This S44 step prioritizes allocating the load to devices with low attenuation rates, which can reduce the additional losses of devices with high attenuation rates and slow down their performance degradation. At the same time, it controls the load increment to not exceed the optimal upper limit, avoiding equipment failures caused by sudden load increases. Combined with regular review and optimization of the allocation rules, it can further adapt to changes in equipment status, reduce the frequency of equipment failures and maintenance needs, extend the stable operation cycle, and reduce equipment procurement and maintenance costs.
[0042] In addition, this solution can also build a dedicated performance degradation curve model for each device based on preprocessed historical data. The model reflects the changing trend of key parameters of the device over time under the current load. By analyzing the vibration amplitude change of a device under the current load over the past 3 months, an amplitude rise curve for the next 1-2 months is fitted, intuitively presenting the performance degradation rhythm of the device. Combining the current actual load of the device with the degradation curve model, the performance degradation rate of each device under the current load is calculated, focusing on evaluating the estimated time for core parameters to drop from the current state to the threshold of the next level state. The predicted degradation curve is compared with the real-time operating parameters of the device. If the deviation between the real-time parameter changes and the predicted trend exceeds a reasonable range, the degradation curve model is corrected based on the latest data.
[0043] By analyzing the changes in core parameters (such as vibration amplitude) of equipment over the past 3 months and fitting the decay curve, the performance decay rate in the next 1-2 months can be presented intuitively. It can also accurately calculate the estimated time for core parameters to drop to the threshold of the next level of state. This allows maintenance to shift from post-fault repair to pre-fault prediction. For example, if it is known in advance that the vibration amplitude of a certain piece of equipment will reach the warning threshold in 1 month, maintenance can be arranged in advance to avoid production interruption caused by sudden failure. It provides data support for load distribution, avoids operation with defects, and can clearly distinguish the anti-degradation ability of different devices by combining the current load of the device with the attenuation rate calculated by the attenuation curve model. Devices with low attenuation rate can bear more load, while devices with high attenuation rate need to reduce the load to delay degradation. This distribution logic based on the device's specific state avoids blindly allocating load due to a lack of understanding of the device's attenuation status, prevents high attenuation rate devices from operating with defects, and reduces excessive wear of components.
[0044] Please see Figure 5 The diagram below illustrates a semiconductor packaging test optimization system according to the present invention. The present invention also provides a semiconductor packaging test optimization system for implementing the aforementioned semiconductor packaging test optimization method. Specifically, the system may include a parameter acquisition module 100, a preference value calculation module 200, a device partitioning module 300, and a device adjustment module 400. More specifically: The parameter acquisition module 100 is used to acquire the operating parameters of the current packaging and testing equipment in real time based on sensors. The preference value calculation module 200 is used to preprocess and analyze the acquired operating parameters, determine the weight values of different indicators, and calculate the current state preference value of the packaging and testing equipment based on the weight values of the indicators. The device partitioning module 300 is used to construct a state preference value range based on historical packaging and testing equipment data, and to substitute the obtained state preference values into different ranges. The equipment adjustment module 400 is used to dynamically adjust the semiconductor loading amount of the packaging and testing equipment according to the state preference value range in which the packaging and testing equipment is located.
[0045] The above description merely illustrates preferred embodiments of the present invention and is quite specific and detailed; however, it should not be construed as limiting the scope of the invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the inventive concept, and these all fall within the scope of protection of the present invention. Therefore, the scope of protection of this invention should be determined by the appended claims.
Claims
1. A semiconductor package test optimization method, the method comprising the steps of: S1: acquiring real-time operation parameters of a current package test device based on sensors; S2: preprocessing and analyzing the acquired operation parameters to determine weight values of different indicators, and calculating a state preference value of the current package test device according to the weight values of the indicators; S3: constructing a state preference value interval according to historical package test device data, and substituting the obtained state preference value into different intervals; S4: dynamically adjusting the semiconductor loading amount of the package test device according to the state preference value interval in which the package test device is located.
2. The method of semiconductor package testing optimization of claim 1, wherein, The sensors include a laser displacement sensor, a torque sensor, a vibration sensor, a temperature sensor, and a current sensor; the laser displacement sensor is disposed at a positioning position of a wafer table and a package mold, and is used to detect positioning and offset parameters; the torque sensor is disposed at an output shaft of a driving motor, and is used to detect speed and rotational speed parameters; the vibration sensor is disposed at a device rack, a motor shell, and a guide rail base, and is used to detect stress and vibration parameters; the temperature sensor is disposed at a welding area, and is used to detect temperature parameters; and the current sensor is disposed at a device total power supply circuit, and is used to detect power consumption parameters.
3. The method of semiconductor package testing optimization of claim 1, wherein, The preprocessing and analysis of the acquired operation parameters include the following sub-steps: S21: data cleaning, outlier rejection, and data standardization processing of the operation parameters; S22: constructing three-level evaluation indicators of a target layer, a criterion layer, and an indicator layer based on the preprocessed operation parameters; wherein the target layer is a device comprehensive state evaluation, the criterion layer is a mechanical system, an environmental process, an electrical performance, and a special process dimension, and the indicator layer is specific parameters under each criterion layer.
4. The method of claim 3, wherein the test optimization is performed for each of the plurality of semiconductor packages. The determination of the weight values of different indicators includes the following sub-steps: S23: determining the weight values by using an analytic hierarchy process, constructing a judgment matrix, inviting experts in the field of semiconductor package testing, scoring the importance of the criterion layer and the indicator layer based on a 1-9 scale method, and forming the judgment matrix; S24: calculating a consistency ratio CR of the judgment matrix, and if CR < 0.1, the test is passed, and if the test is not passed, the experts are fed back to adjust the judgment matrix until the test is passed; S25: calculating a characteristic vector corresponding to the maximum eigenvalue of the judgment matrix by using an eigenvalue method, and obtaining a criterion layer subjective weight and an indicator layer subjective weight; S26: calculating an objective weight according to historical evaluation data: ； ； wherein, represents the proportion of the standardized value of the jth cycle and the ith index, and n represents the number of cycles, represents the entropy value of the ith evaluation index, and ln represents the natural logarithm operation, represents the objective weight of the ith index; S27: fusing the subjective weight and the objective weight at a proportion of 7:3 to obtain a final weight: ； wherein, represents the maximum weight of the i-th indicator, represents the subjective weight of the i-th indicator.
5. The semiconductor packaging testing optimization method as described in claim 1, characterized in that, The specific calculation process of calculating the state preference value of the current package test device according to the weight values of the indicators is as follows: ； wherein, represents a state preference value of the kth criterion layer, represents the number of indicators of the kth criterion layer, represents a normalized value of the ith indicator of the kth criterion layer, represents the maximum weight of the ith indicator.
6. The method of claim 1, wherein, The S3 step includes the following sub-steps: S31: dividing intervals according to percentiles based on different device states in historical package test device data, taking P90, P60, and P40 as boundaries, and the historical package test device data includes package test device data in normal working states, pre-warning states, and fault states; S32: substituting the calculated state preference value into different intervals to divide and judge the state of the current package test device.
7. The method of semiconductor package testing optimization of claim 1, wherein, The S4 step includes the following sub-steps: S41: According to the interval of different packaging test equipment, the corresponding control instruction is generated, and the control instruction is issued to the corresponding packaging test equipment, and the semiconductor feeding amount of the packaging test equipment is adjusted; S42: Repeat steps S1-S4 within a predetermined interval to re-adjust the semiconductor feeding amount of the packaging test equipment.
8. The method of semiconductor package testing optimization of claim 7, wherein, Before the S41 step, the following substeps are further included: S40: A mapping relationship library of the state interval of the packaging test equipment and the semiconductor feeding amount parameter is established in advance.
9. The method of semiconductor package testing optimization of claim 1, wherein, The S4 step further includes the following substeps: S43: According to the running data of the whole life cycle of the packaging test equipment, a performance attenuation curve model is constructed; S44: Calculate the attenuation rate of each packaging test equipment under the current load, and allocate the load for each packaging test equipment according to the attenuation rate and the balance coefficient.
10. A semiconductor package test optimization system, comprising: The system includes the following modules: A parameter acquisition module for acquiring the running parameters of the current packaging test equipment in real time based on the sensor; A preference value calculation module for analyzing the preprocessed running parameters, determining the weight values of different indicators, and calculating the state preference value of the current packaging test equipment according to the weight values of the indicators; A device partitioning module for constructing a state preference value interval based on historical packaging test equipment data, and substituting the obtained state preference value into different intervals; A device adjustment module for dynamically adjusting the semiconductor feeding amount of the packaging test equipment according to the state preference value interval of the packaging test equipment.