Vertical quantitative analysis method for chip mass production test by adopting big data analysis mode

Abstract

The invention relates to the technical field of electronic testing, in particular to a vertical quantitative analysis method for chip mass production testing by adopting a big data analysis mode, which comprises the following steps: collecting time and error rate data of the chip mass production testing, carrying out sorting and clustering analysis on the data by adopting the big data analysis mode, identifying key distribution and abnormal points in the data, and carrying out vertical quantitative analysis on the chip mass production testing. And obtaining a time distribution analysis result. According to the method, segmentation and vertical quantitative analysis are performed on time and error rate data through a big data technology, abnormal data points exceeding a preset threshold value can be marked and rapidly identified, and the identification capability of potential problems in the production process is improved by comparing chip performances under different production batches and process parameters; through quantitative scoring and risk difference analysis between batches, the risk management and response capability is further enhanced, powerful support is provided for optimizing the chip mass production test process and parameter setting, and the test efficiency is greatly improved.

Description

Technical Field

[0001] The present invention relates to the field of electronic testing technology, and in particular to a vertical quantitative analysis method for chip mass production testing using a big data analysis method. Background Art

[0002] The field of electronic testing technology mainly involves the use of electronic equipment and instruments to detect and verify the functions, performance and reliability of electronic equipment or components. This field includes a series of technologies and methods, such as circuit analysis, signal integrity testing, functional verification, fault diagnosis and performance evaluation. In the semiconductor manufacturing process, electronic testing technology is particularly important to ensure that the chip can meet the design requirements before leaving the factory. With the advancement of technology and the increase in complexity, electronic testing is constantly incorporating advanced algorithms and big data analysis technologies to improve the efficiency and accuracy of testing.

[0003] Among them, the vertical quantitative analysis method of chip mass production testing using big data analysis refers to the use of big data analysis technology to conduct in-depth data processing and analysis by collecting batch production and test data during the mass production stage of the chip. The purpose of this method is to improve the accuracy and efficiency of testing through data analysis, quickly identify anomalies and trends in the production process, optimize the production process and improve product quality. Through vertical quantitative analysis, the performance of chips under different production batches and different process parameters can be systematically compared to help companies make more accurate production decisions.

[0004] Although existing electronic testing technologies are capable of basic functional verification and performance evaluation, they lack sufficient in-depth analysis capabilities when processing large-scale production data. Especially in the semiconductor industry, the complexity of mass production and the increasing sophistication of chip design require more efficient data processing methods. In traditional methods, a single testing technology is difficult to quickly and accurately identify anomalies and trends in the production process, resulting in delayed or incorrect production decisions, increased product failure rates and production costs. Existing technologies are also limited in their ability to manage risks and optimize production processes. This makes it impossible for companies to effectively respond to sudden production problems and changes in market demand in a dynamically changing market environment, affecting their competitiveness and market response speed. Summary of the invention

[0005] In order to solve the problem that the existing technology lacks sufficient in-depth analysis capabilities when processing large-scale production data. Especially in the semiconductor industry, the complexity of mass production and the increasing sophistication of chip design require more efficient data processing methods. In traditional methods, a single testing technology is difficult to quickly and accurately identify anomalies and trends in the production process, resulting in delayed or incorrect production decisions, increased product failure rates and production costs. The ability of existing technologies in risk management and optimization of production processes is also relatively limited. This leads to the inability of enterprises to effectively respond to sudden production problems and changes in market demand in a dynamically changing market environment, affecting the competitiveness of enterprises and the speed of market response. The technical problem, the embodiment of the present invention provides a vertical quantitative analysis method for chip mass production testing using big data analysis. The technical solution is as follows:

[0006] On the one hand, a vertical quantitative analysis method for chip mass production testing using big data analysis is provided, and the method includes:

[0007] S1: Collect the time and error rate data of chip mass production test, use big data analysis to sort and cluster the data, identify the key distribution and abnormal points in the data, and obtain the time distribution analysis results;

[0008] S2: using the time distribution analysis result, calculating the chip mass production test parameters in the future time period through a linear regression model, and comparing with the real-time data, predicting the change trend of the chip mass production parameters, and obtaining the parameter change prediction result;

[0009] S3: Based on the parameter change prediction results, the time and error rate data of chip mass production test are segmented by big data analysis, the volatility within multiple segments is vertically quantified and analyzed, and data points exceeding the preset threshold are marked to obtain anomaly detection results;

[0010] S4: Based on the abnormal detection results, vertically quantify and analyze the error rate factors of chip mass production testing, compare the abnormal data with the preset failure mode, identify the key factors causing chip failure, and obtain the failure factor analysis results;

[0011] S5: Using the failure factor analysis results, quantitatively score the risks of the chip mass production test batches, analyze the risk differences between batches, take risk measures, and obtain risk assessment results;

[0012] S6: Based on the risk assessment results, optimize the chip mass production test process and parameter settings, optimize the test efficiency by dynamically adjusting the test frequency and resource allocation, and obtain vertical quantitative analysis results.

[0013] As a further scheme of the present invention, the time distribution analysis results include a time series graph of the data, the time of the abnormal points, and the key statistical information of the data; the parameter change prediction results include the predicted parameter change curve, the change range of the associated parameters, and the prediction consistency evaluation; the anomaly detection results include the identified abnormal time period, the abnormality degree assessment, and the influencing factors of the abnormal points; the failure factor analysis results include failure mode identification, key influencing factors, and weight scores of influencing factors; the risk assessment results include risk level classification, key risk factors, and the cumulative impact of differentiated risk factors; the vertical quantitative analysis results include the adjusted resource allocation plan, test interval time, and expected efficiency optimization indicators.

[0014] As a further solution of the present invention, the time and error rate data of chip mass production tests are collected, and the data are sorted and clustered by big data analysis to identify key distributions and abnormal points in the data. The specific steps of obtaining the time distribution analysis results are as follows:

[0015] S101: Collect time data and error rate data in chip mass production testing, organize the data, remove incomplete and damaged records, screen the data and exclude outliers, and obtain a processed data set;

[0016] S102: sorting the time and error rate data of chip mass production test according to the processed data set, using time series sorting to arrange the data from early to late according to the test date, and performing secondary sorting according to the error rate, analyzing the distribution and trend of the data on the time axis, and obtaining the sorted data distribution result;

[0017] S103: Based on the sorted data distribution result, the K-means clustering algorithm is used to divide the data into multiple clusters, calculate the distance from the data point to the cluster center, identify the key distribution and abnormal points in the cluster, and obtain the time distribution analysis result.

[0018] As a further solution of the present invention, the formula for calculating the distance from the data point to the cluster center is:

[0019]

[0020] Among them, A i represents the distance from the i-th data point to the cluster center, x ij represents the value of the i-th data point in the j-th dimension, μ j represents the mean of the cluster center in the jth dimension, σ j represents the standard deviation of the jth dimension, w j represents the weight coefficient of the jth dimension, and k is the number of dimensions.

[0021] As a further solution of the present invention, the time distribution analysis results are used to calculate the chip mass production test parameters in the future time period through a linear regression model, and the real-time data is compared to predict the change trend of the chip mass production parameters. The steps of obtaining the parameter change prediction results are specifically as follows:

[0022] S201: using the time distribution analysis result, selecting variables and data sets for model construction, setting model parameters, including slope and intercept, and using real-time data for model training to construct a linear regression model;

[0023] S202: Based on the linear regression model, calculate and predict the chip mass production test parameters in the future time period, evaluate the prediction results using the data inferred by the linear regression model, verify the prediction ability of the linear regression model, and obtain the prediction calculation results;

[0024] S203: Based on the prediction calculation results, the predicted chip mass production parameters are compared and analyzed with the real-time data to evaluate the differences and trends between the two, predict the change trends of the chip mass production parameters in the future time period, and obtain parameter change prediction results.

[0025] As a further solution of the present invention, the parameter change prediction results are used to segment the time and error rate data of chip mass production tests by adopting a big data analysis method, vertically quantify and analyze the volatility within multiple segments, mark the data points that exceed the preset threshold, and obtain the abnormal detection results in the following steps:

[0026] S301: using the parameter change prediction result and big data analysis technology, dividing the time and error rate data of chip mass production test into multiple time periods, analyzing the volatility and trend of the data within the target time range, and obtaining data segmentation results;

[0027] S302: Based on the data segmentation results, vertical quantitative analysis is performed on the data within the time period, and statistical indicators of the data within multiple segments are calculated, including the mean value, standard deviation, and coefficient of variation, to identify key features of data volatility and obtain volatility analysis results;

[0028] S303: Using the volatility analysis result, set a preset threshold, mark the data points that exceed the preset threshold, analyze potential production problems and chip mass production test anomalies, and obtain anomaly detection results.

[0029] As a further solution of the present invention, based on the abnormal detection results, vertical quantitative analysis of the error rate factors of chip mass production testing is performed, the abnormal data is compared with the preset failure mode, and the key factors causing chip failure are identified. The steps of obtaining the failure factor analysis results are specifically as follows:

[0030] S401: Based on the abnormal detection result, screen and extract the identified abnormal data of chip mass production test, separate the data points beyond the normal range, and generate abnormal data screening results;

[0031] S402: Based on the abnormal data screening result, the screened abnormal data points are compared with the preset failure modes, the correlation between the data points and the failure modes is checked, and the comparative analysis results are obtained;

[0032] S403: Using the comparative analysis results, identify key error rate factors behind the abnormal data points, analyze the impact of the key error rate factors on chip failure, and obtain failure factor analysis results.

[0033] As a further solution of the present invention, the failure factor analysis results are used to quantitatively score the risks of chip mass production test batches, analyze the risk differences between batches, and take risk measures. The specific steps for obtaining the risk assessment results are:

[0034] S501: Based on the failure factor analysis results, select key failure factors as risk assessment indicators, assign risk scores to chips in the test batches, analyze risk levels of multiple batches, and obtain risk score records;

[0035] S502: using the risk scoring records, analyzing the risk score differences between differentiated batches, identifying risky batches, risk sources and risk distribution, and obtaining batch risk analysis results;

[0036] S503: Based on the batch risk analysis results, formulate targeted risk response measures, including adjusting production parameters, strengthening quality control and preventive maintenance, avoiding the probability of risk occurrence in future time periods, and obtaining risk assessment results.

[0037] As a further solution of the present invention, based on the risk assessment results, the chip mass production test process and parameter settings are optimized, and the test efficiency is optimized by dynamically adjusting the test frequency and resource allocation. The steps of obtaining the vertical quantitative analysis results are specifically as follows:

[0038] S601: Based on the risk assessment result, analyze the risk factors in the batch, adjust the key parameter settings in the chip mass production test process according to the risk factors, including temperature, pressure and voltage, and obtain parameter adjustment results;

[0039] S602: using the parameter adjustment result, adjusting the chip mass production test cycle and frequency according to the differentiated risk levels, calculating the optimized test cycle, optimizing the use efficiency of test resources, and obtaining a test frequency adjustment record;

[0040] S603: Utilize the test frequency adjustment record to optimize the efficiency of the chip mass production test process, evaluate the performance of the optimized test process in reducing risks and optimizing chip yield, and obtain vertical quantitative analysis results.

[0041] As a further solution of the present invention, the formula for calculating the optimized test period is:

[0042]

[0043] Among them, f(x) is the optimized test cycle, R represents the risk level, D represents the number of chips to be tested, P represents the number of available test platforms, T represents the average test time, and a, b and c are weight parameters.

[0044] The beneficial effects brought about by the technical solution provided by the embodiment of the present invention include at least:

[0045] By collecting and analyzing batch production test data, the accuracy and efficiency of chip mass production testing can be improved. By sorting and clustering the data, the key distribution and abnormal points in the data can be effectively identified, and the trend of future chip mass production test parameter changes can be accurately predicted. By using big data technology to segment and vertically quantify time and error rate data, abnormal data points that exceed the preset threshold can be marked and quickly identified. By comparing chip performance under different production batches and process parameters, the ability to identify potential problems in the production process can be improved, allowing companies to optimize production processes and improve product quality in a targeted manner. By quantitatively scoring and analyzing risk differences between batches, the ability to manage and respond to risks can be further enhanced, providing strong support for optimizing chip mass production test processes and parameter settings, greatly improving test efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 It is a schematic diagram of the workflow of the present invention;

[0047] Figure 2 This is a detailed flow chart of S1 of the present invention;

[0048] Figure 3 This is a detailed flow chart of S2 of the present invention;

[0049] Figure 4 This is a detailed flow chart of S3 of the present invention;

[0050] Figure 5 This is a detailed flow chart of S4 of the present invention;

[0051] Figure 6 This is a detailed flow chart of S5 of the present invention;

[0052] Figure 7 This is a detailed flow chart of S6 of the present invention. DETAILED DESCRIPTION

[0053] The technical solution of the present invention is described below in conjunction with the accompanying drawings.

[0054] In the embodiments of the present invention, words such as "exemplarily" and "for example" are used to indicate examples, illustrations or explanations. Any embodiment or design described as "exemplary" in the present invention should not be interpreted as being more preferred or more advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present concepts in a specific way. In the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either of the two.

[0055] In order to make the technical problems, technical solutions and advantages to be solved by the present invention more clear, a detailed description will be given below with reference to the accompanying drawings and specific embodiments.

[0056] See also Figure 1 The embodiment of the present invention provides a vertical quantitative analysis method for chip mass production testing using big data analysis. The processing flow of the method may include the following steps:

[0057] S1: Collect the time and error rate data of chip mass production test, use big data analysis to sort and cluster the data, identify the key distribution and abnormal points in the data, and obtain the time distribution analysis results;

[0058] S2: Using the time distribution analysis results, the linear regression model is used to calculate the chip mass production test parameters in the future time period, and the real-time data is compared to predict the change trend of the chip mass production parameters to obtain the parameter change prediction results;

[0059] S3: Based on the parameter change prediction results, the time and error rate data of chip mass production test are segmented by big data analysis, and the volatility within multiple segments is vertically quantified and analyzed. Data points that exceed the preset threshold are marked to obtain anomaly detection results.

[0060] S4: Based on the abnormal detection results, vertically quantify and analyze the error rate factors of chip mass production testing, compare the abnormal data with the preset failure mode, identify the key factors that cause chip failure, and obtain the failure factor analysis results;

[0061] S5: Using the failure factor analysis results, quantitatively score the risks of the chip mass production test batches, analyze the risk differences between batches, take risk measures, and obtain risk assessment results;

[0062] S6: Based on the risk assessment results, optimize the chip mass production test process and parameter settings, optimize the test efficiency by dynamically adjusting the test frequency and resource allocation, and obtain vertical quantitative analysis results.

[0063] The results of time distribution analysis include time series graph of data, time of abnormal points, and key statistical information of data. The results of parameter change prediction include predicted parameter change curve, change range of associated parameters, and prediction consistency evaluation. The results of anomaly detection include identified abnormal time period, abnormal degree evaluation, and influencing factors of abnormal points. The results of failure factor analysis include failure mode identification, key influencing factors, and weight scores of influencing factors. The results of risk assessment include risk level classification, key risk factors, and the cumulative impact of differentiated risk factors. The results of vertical quantitative analysis include adjusted resource allocation plan, test interval time, and expected efficiency optimization indicators.

[0064] See also Figure 2 , collect the time and error rate data of chip mass production test, use big data analysis to sort and cluster the data, identify the key distribution and abnormal points in the data, and obtain the time distribution analysis results in the following steps:

[0065] S101: Collect time data and error rate data in chip mass production test, sort the data, remove incomplete and damaged records, filter the data and exclude outliers, and obtain the processed data set. The execution process is as follows;

[0066] The process of collecting time data and error rate data in chip mass production testing includes multiple detailed steps. It is the initial collection of data, including automatically recording the test time and error rate of the chip from the test equipment on the production line, organizing the initially collected data, and removing incomplete or damaged records. This involves data verification and data cleaning techniques to ensure the data quality of subsequent analysis, and data screening to exclude outliers that deviate significantly from the normal range. This step is achieved by setting thresholds for error rate and time data for further analysis and evaluation to help understand the time efficiency and product quality issues in the production process, provide support for improving the test process and improving product quality, and obtain a processed data set.

[0067] S102: sorting the time and error rate data of chip mass production test according to the processed data set, using time series sorting to arrange the data from early to late according to the test date, and performing secondary sorting according to the error rate, analyzing the distribution and trend of the data on the time axis, and obtaining the sorted data distribution result. The execution process is as follows;

[0068] The steps to sort the time and error rate data of chip mass production testing include using time series sorting technology to arrange the data from early to late according to the test date, which not only helps to understand the distribution of data on the time axis, but also makes it easy to identify trends and patterns in the data set through this sorting. The secondary sorting of error rate data is carried out from low to high, which helps to quickly locate batches or individual chips with poor performance. Such sorting not only makes the data more orderly, but also facilitates subsequent statistical analysis and the formulation of quality control measures. By analyzing the distribution of sorted data, we can better understand the relationship between time and error rate in the chip testing process, make corresponding adjustments in production strategies to improve production efficiency and reduce error rates, and obtain the sorted data distribution results.

[0069] S103: Based on the sorted data distribution results, the K-means clustering algorithm is used to divide the data into multiple clusters, calculate the distance from the data point to the cluster center, identify the key distribution and abnormal points in the cluster, and obtain the time distribution analysis results. The execution process is as follows;

[0070] The formula for calculating the distance from a data point to the cluster center is:

[0071]

[0072] Among them, A i represents the distance from the i-th data point to the cluster center, x ij represents the value of the i-th data point in the j-th dimension, μ j represents the mean of the cluster center in the jth dimension, σ j represents the standard deviation of the jth dimension, w j represents the weight coefficient of the jth dimension, and k is the number of dimensions;

[0073] Parameter meaning and setting value:

[0074] x ij is the value of the i-th data point in the j-th dimension. Suppose there is a data point whose values ​​in the two dimensions are 4.5 and 3.2 respectively;

[0075] μ j is the mean of the cluster center in the jth dimension, and the mean of the cluster centers in the two dimensions are set to 5.0 and 3.0 respectively;

[0076] σ j is the standard deviation of the jth dimension, ensuring that the distance calculation has a standardized comparison benchmark in each dimension, and the standard deviations are set to 0.5 and 0.8 respectively;

[0077] w jis the weight coefficient of the jth dimension, indicating the relative importance of each dimension in cluster analysis. The weight coefficients are set to 1.0 and 1.5 respectively. According to the previous statistical analysis, the weight reflects the contribution of the dimension in distinguishing different clusters.

[0078] Substitute the parameters into the formula for calculation:

[0079]

[0080] The results show that the distance between the data point and the cluster center is 1.046. This value reflects the relative position of the data point and the cluster center. A low distance value indicates that the data point is closer to the cluster center and has a higher probability of being classified into the cluster.

[0081] See also Figure 3 , using the time distribution analysis results, through the linear regression model, calculate the chip mass production test parameters in the future time period, and compare with the real-time data to predict the change trend of the chip mass production parameters. The specific steps to obtain the parameter change prediction results are as follows:

[0082] S201: Using the time distribution analysis results, select variables and data sets for model construction, set model parameters, including slope and intercept, and use real-time data for model training. The execution process of building a linear regression model is as follows;

[0083] In the process of model building, it is necessary to select appropriate variables and data sets. This process requires a comprehensive analysis of past data to determine factors that have a direct impact on chip mass production efficiency and quality, such as temperature, humidity, and machine speed. The data selection and processing stage includes data cleaning, outlier removal, and data standardization to ensure the quality of data input during model training. The setting of model parameters, such as slope and intercept, is also a key step in building the model. The parameters are obtained through historical data analysis and prediction experiments. Adjusting the slope of the model needs to be optimized by minimizing the sum of squared errors based on the results of the preliminary model test, while the setting of the intercept directly affects the deviation of the model, which can more accurately reflect the data laws in actual production, play a role in practical applications, improve the efficiency of chip production and reduce costs, and build a linear regression model.

[0084] S202: Based on the linear regression model, the chip mass production test parameters in the future time period are calculated and predicted, and the prediction results are evaluated using the data inferred by the linear regression model to verify the prediction ability of the linear regression model. The execution process of obtaining the prediction calculation results is as follows;

[0085] The prediction results are evaluated using the data derived from the linear regression model according to the formula:

[0086] Y = βX + ∈;

[0087] Calculate the predicted value Y of the chip mass production test parameter, where β represents the coefficient of the variable, X represents the independent variable, and ∈ represents the error term;

[0088] Set the real-time collected data to display a chip test parameter X=200;

[0089] The coefficient β determined through historical data analysis is 1.2;

[0090] The error term ∈ is set to 5;

[0091] Substituting the parameters into the formula, we can calculate:

[0092] Y = 1.2 × 200 + 5 = 245;

[0093] The result shows that the predicted chip mass production test parameter is 245. The predicted value will be used to evaluate the accuracy of the model and further compared with the test parameters in actual production to verify the predictive ability of the linear regression model.

[0094] S203: Based on the prediction calculation results, the predicted chip mass production parameters are compared and analyzed with the real-time data to evaluate the differences and trends between the two, and the change trends of the chip mass production parameters in the future time period are predicted. The execution process of obtaining the parameter change prediction results is as follows;

[0095] The comparative analysis between the predicted calculation results and the real-time data is a complex process, which involves not only the direct comparison of data, but also the in-depth analysis of the differences between the two, as well as the trend prediction based on the differences. It requires the collection and organization of predicted data and real-time data to ensure that the two sets of data are comparable in time and parameters. Statistical analysis methods, such as regression analysis and variance analysis, are used to quantitatively describe the differences between the two, find out the estimated anomalies and causes of data fluctuations, and perform trend analysis on the differences. For example, the future trend of changes can be predicted through moving average or exponential smoothing methods. The analysis results are used to improve the parameter settings for chip mass production, optimize the production efficiency and quality of the chip by adjusting the key variables in the production process, ensure the stability and reliability of the chip mass production process, and obtain parameter change prediction results.

[0096] See also Figure 4 ,Through the parameter change prediction results, the big data analysis method is adopted,to segment the time and error rate data of chip mass production test, vertically quantified the volatility within multiple segments, and the data points exceeding the preset threshold are marked.,The specific steps to obtain the anomaly detection results are:

[0097] S301: Using the parameter change prediction results and big data analysis technology, the time and error rate data of chip mass production test are divided into multiple time periods, and the volatility and trend of the data within the target time range are analyzed to obtain the data segmentation results. The execution process is as follows;

[0098] When using big data analysis technology to analyze the time and error rate data of chip mass production tests, the data is divided into multiple time periods. This process involves time series decomposition of the data, as well as preliminary screening and classification of error rate data. Trend analysis is performed on the data in each time period. By calculating volatility indicators of each segment, such as volatility and fluctuation range, the stability and abnormalities of the data are evaluated. This helps to identify problems or incidents in the production process and provide a basis for subsequent analysis. The data will be used for further quality control and process optimization to obtain data segmentation results.

[0099] S302: Based on the data segmentation results, vertical quantitative analysis is performed on the data within the time period, and statistical indicators of the data within multiple segments are calculated, including the mean, standard deviation, and coefficient of variation, to identify key features of data volatility. The execution process of obtaining the volatility analysis results is as follows;

[0100] The process of vertical quantitative analysis of data within a time period includes several key steps. Statistical indicators such as mean, standard deviation and coefficient of variation are calculated for each time period. The indicators help to quantitatively describe the performance and volatility of data in different time periods. Through statistical indicators, key characteristics of data volatility can be identified. For example, by comparing the standard deviations of different time periods, it can be determined that there are potential stability problems in the production process of the time period. The calculation of the coefficient of variation provides a relative volatility measure, which helps to compare between production batches of different sizes. The statistical analysis results will be used for further quality control and adjustment of the production process to obtain volatility analysis results.

[0101] S303: Using the volatility analysis results, setting a preset threshold, marking the data points that exceed the preset threshold, analyzing potential production problems and chip mass production test anomalies, and obtaining the execution process of the anomaly detection results are as follows;

[0102] Using the volatility analysis results, set a preset threshold and mark the data points that exceed the preset threshold according to the formula:

[0103] T = μ + Bσ;

[0104] In the formula, T is the preset threshold, μ represents the mean value, σ represents the standard deviation, and B represents the multiple of the standard deviation;

[0105] Set the mean value μ of a certain time period obtained from the volatility analysis to 50, the standard deviation σ to 5, and use 3 times the standard deviation as the threshold setting;

[0106] Substitute the parameters into the formula for calculation:

[0107] T = 50 + 3 × 5 = 65;

[0108] Data points that exceed this threshold are marked as anomalies, indicating that the data points represent anomalies in production or deviations from normal production parameters. This marking helps to quickly identify potential problem areas that require further analysis and investigation, and take appropriate improvement measures.

[0109] See also Figure 5 Based on the abnormal detection results, vertical quantitative analysis of the error rate factors of chip mass production testing is carried out, and the abnormal data is compared with the preset failure mode to identify the key factors that cause chip failure. The specific steps to obtain the failure factor analysis results are as follows:

[0110] S401: Based on the abnormal detection results, the identified abnormal data of chip mass production test is screened and extracted, and the data points beyond the normal range are separated. The execution process of generating the abnormal data screening results is as follows;

[0111] The steps of screening and extracting the identified abnormal data of chip mass production tests involve complex data processing technology, which is to screen out data points that exceed the normal performance range from batch production data. This process uses automated data screening tools to quickly identify and isolate data points with abnormal values, such as excessive temperature or unstable voltage, and classify and organize the data to separate the truly abnormal data, that is, not only the abnormal values, but also the key data points that affect chip performance. This will provide a clear basis for further analysis and processing, ensure the accuracy and efficiency of subsequent steps, and generate abnormal data screening results.

[0112] S402: Through the abnormal data screening results, the screened abnormal data points are compared with the preset failure modes to check the correlation between the data points and the failure modes, and the execution process of obtaining the comparative analysis results is as follows;

[0113] The process of comparing the screened abnormal data points with the preset failure mode includes detailed data comparison analysis and pattern recognition. The screened abnormal data points need to be classified in detail and matched with the preset failure mode database according to the characteristics of the data, such as amplitude, frequency and duration. The matching process involves machine learning algorithms or statistical analysis methods to ensure that the correlation analysis between data points and failure modes is accurate and reliable. The matched data points are further verified to check whether the data points truly meet the characteristics of the failure mode. This not only helps to confirm the nature of the abnormal data points, but also provides a basis for further processing and preventive measures to obtain comparative analysis results.

[0114] S403: Using the comparative analysis results, identify the key error rate factors behind the abnormal data points, analyze the impact of the key error rate factors on chip failure, and obtain the failure factor analysis results. The execution process is as follows;

[0115] Identify the key error rate factors behind abnormal data points, according to the formula:

[0116] E=∑(p s × s );

[0117] In the formula, E represents the total error rate impact, p s represents the probability of occurrence of each error rate factor s, w s represents the weight of the corresponding factor;

[0118] Set to identify two key error rate factors, with probabilities of occurrence p1 = 0.3 and p2 = 0.7;

[0119] The corresponding influence weights are w1=0.5 and w2=1.5;

[0120] Substitute the parameters into the formula and calculate:

[0121] E=(0.3×0.5)+(0.7×1.5)=0.15+1.05=1.2;

[0122] The calculation results show that, considering the occurrence probability and influence weight of each factor, the critical error rate factor has an impact on the total failure of the chip of 1.2. This value helps to evaluate and quantify the specific role of each failure factor in chip production failure, and provides a quantitative basis for further quality improvement and control measures.

[0123] See also Figure 6 , using the failure factor analysis results, quantitatively score the risks of chip mass production test batches, analyze the risk differences between batches, and take risk response measures. The specific steps to obtain the risk assessment results are as follows:

[0124] S501: Based on the failure factor analysis results, select key failure factors as risk assessment indicators, assign risk scores to the chips in the test batches, analyze the risk levels of multiple batches, and obtain the execution process of risk score records as follows;

[0125] Assign a risk score to the chips in the test batch according to the formula:

[0126] AR=∑(f q ×s q );

[0127] Analyze the risk level of multiple batches, where AR represents the total risk score, f qrepresents the selected critical failure factor, s q Represents the corresponding risk score coefficient;

[0128] For a certain batch, three key failure factors are selected, and the risk scoring coefficients are s1=0.6 and s3=0.1 respectively;

[0129] And the risk contribution of the factor f q f1=2, f3=5 respectively;

[0130] Substituting the parameters into the formula, we can calculate:

[0131] AR=(2×0.6)+(3×0.3)+(5×0.1)=1.2+0.9+0.5=2.6;

[0132] The results show that, considering the risk contribution and scoring coefficient of each key failure factor, the total risk score of the batch is 2.6. This value helps analyze and compare the risk levels between different batches and provides a quantitative basis for subsequent risk management and control strategies.

[0133] S502: Using risk scoring records, analyzing the risk score differences between differentiated batches, identifying risky batches, risk sources and risk distribution, and obtaining batch risk analysis results. The execution process is as follows;

[0134] The process of analyzing the differences in risk scores between differentiated batches involves detailed data statistics and comparative analysis. By collecting and organizing the risk score data of each batch, statistical analysis methods such as variance analysis and cluster analysis are used to quantitatively describe the differences in risk scores between different batches, identify risky batches, risk sources and risk distribution, and reveal through charts and trend lines that some batches have higher risk levels and that the risks mainly come from some failure factors. This will be used to optimize the production process and improve chip quality control standards, reduce the occurrence of future risks, and obtain batch risk analysis results.

[0135] S503: Based on the batch risk analysis results, formulate targeted risk response measures, including adjusting production parameters, strengthening quality control and preventive maintenance, and avoiding the probability of risk occurrence in the future time period. The execution process of obtaining the risk assessment results is as follows;

[0136] The process of developing targeted risk response measures includes several key links. Based on the results of risk analysis, the production parameters that need to be adjusted are determined, including the machine's operating speed, temperature control or raw material usage standards. Quality control is strengthened by introducing stricter testing standards and frequent quality inspections to ensure that the quality of each batch of chips meets the highest standards. Preventive maintenance is implemented, and production equipment is regularly inspected and maintained to prevent production problems caused by equipment failure. The measures will effectively avoid the probability of risk occurrence in the future time period, improve the stability and reliability of the entire production system, and obtain risk assessment results.

[0137] See also Figure 7 Based on the risk assessment results, the chip mass production test process and parameter settings are optimized, and the test efficiency is optimized by dynamically adjusting the test frequency and resource allocation. The specific steps to obtain the vertical quantitative analysis results are as follows:

[0138] S601: Based on the risk assessment results, the risk factors in the batch are analyzed, and the key parameter settings in the chip mass production test process are adjusted according to the risk factors, including temperature, pressure and voltage. The execution process of obtaining the parameter adjustment results is as follows;

[0139] The process of analyzing risk factors in a batch involves detailed data statistics and risk identification. Through in-depth comparison of historical data and current batch data, key factors that lead to production risks, such as temperature, pressure, and voltage, are determined. According to the risk factors, the key parameters in the chip mass production test process are adjusted. The adjustment process requires precise control of each parameter to ensure that the optimization standards are met in terms of safety and efficiency. The process also includes real-time monitoring and evaluation of the effect of parameter adjustment to ensure that the adjustment achieves the expected risk control effect, which will directly affect the production quality and output of the chip. Through adjustment, the risks in production can be significantly reduced, the overall performance and output of the chip can be improved, and the parameter adjustment results can be obtained.

[0140] S602: Using the parameter adjustment result, adjusting the chip mass production test cycle and frequency according to the differentiated risk level, calculating the optimized test cycle, optimizing the use efficiency of test resources, and obtaining the execution process of the test frequency adjustment record is as follows;

[0141] The formula for calculating the optimized test period is:

[0142]

[0143] Where f(x) is the optimized test cycle, R represents the risk level, D represents the number of chips to be tested, P represents the number of available test platforms, T represents the average test time, and a, b, and c are weight parameters;

[0144] Parameter meaning and setting value:

[0145] R is the risk level, which is calculated by the risk assessment model. The value range here is 1 to 10. The specific value depends on the failure rate of the chip and historical test data.

[0146] D is the number of chips to be tested, which is obtained through real-time monitoring of production line data, and an example is 500 chips;

[0147] P is the number of available test platforms, which is obtained based on the data provided by the equipment management system. In this example, 20 are taken;

[0148] T is the average test time, which is calculated by averaging historical test data and is set to 15 hours;

[0149] a, b, and c are the weight parameters of risk level, resource utilization, and time, respectively, which are adjusted according to historical optimization results, and c = 0.2;

[0150] Substitute the parameters into the formula for calculation:

[0151]

[0152] The result 12.274 shows that the optimized test cycle is 12.274 units of time, which means that according to the current resource and risk situation, the test cycle is adjusted compared with the standard situation to adapt to the current production needs and risk assessment. The determination of this cycle provides a basis for chip mass production testing, making resource utilization and risk management more reasonable, and further optimizing the test process and efficiency.

[0153] S603: Utilize the test frequency adjustment record to optimize the efficiency of the chip mass production test process, evaluate the performance of the optimized test process in reducing risks and optimizing chip yield, and obtain the execution process of the vertical quantitative analysis result as follows;

[0154] The steps to optimize the efficiency of the chip mass production test process include evaluating the rationality and efficiency of the current test frequency, and determining whether the test frequency needs to be adjusted to adapt to the production rhythm and quality requirements by analyzing the data collected during the test process. This involves adjusting the settings of the test equipment and reconfiguring the test process to ensure that the tests can effectively capture the estimated defects and deficiencies, and evaluating the performance of the optimized test process in reducing risks and optimizing chip yields. The evaluation includes statistical analysis of the test data, such as analyzing the defect rate, yield changes and production efficiency improvements after testing. Through the data, the specific impact of the test frequency adjustment on the production process can be clearly seen, and vertical quantitative analysis results can be obtained.

[0155] The above is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed by the present invention, which should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A vertical quantitative analysis method for chip mass production testing using big data analysis, characterized in that: The following steps are involved: S1: Collect the time and error rate data of chip mass production test, use big data analysis to sort and cluster the data, identify the key distribution and abnormal points in the data, and obtain the time distribution analysis results; S2: using the time distribution analysis result, calculating the chip mass production test parameters in the future time period through a linear regression model, and comparing with the real-time data, predicting the change trend of the chip mass production parameters, and obtaining the parameter change prediction result; S3: Based on the parameter change prediction results, the time and error rate data of chip mass production test are segmented by big data analysis, the volatility within multiple segments is vertically quantified and analyzed, and data points exceeding the preset threshold are marked to obtain anomaly detection results; S4: Based on the abnormal detection results, vertically quantify and analyze the error rate factors of chip mass production testing, compare the abnormal data with the preset failure mode, identify the key factors causing chip failure, and obtain the failure factor analysis results; S5: Using the failure factor analysis results, quantitatively score the risks of chip mass production test batches, analyze the risk differences between batches, take risk measures, and obtain risk assessment results; S6: Based on the risk assessment results, optimize the chip mass production test process and parameter settings, optimize the test efficiency by dynamically adjusting the test frequency and resource allocation, and obtain vertical quantitative analysis results.

2. The vertical quantitative analysis method for chip mass production testing using big data analysis according to claim 1 is characterized in that: The time distribution analysis results include a time series graph of the data, the time of the abnormal points, and the key statistical information of the data. The parameter change prediction results include the predicted parameter change curve, the change range of the associated parameters, and the prediction consistency evaluation. The anomaly detection results include the identified abnormal time period, the abnormal degree assessment, and the influencing factors of the abnormal points. The failure factor analysis results include failure mode identification, key influencing factors, and weight scores of influencing factors. The risk assessment results include risk level classification, key risk factors, and the cumulative impact of differentiated risk factors. The vertical quantitative analysis results include the adjusted resource allocation plan, test interval time, and expected efficiency optimization indicators.

3. The vertical quantitative analysis method for chip mass production testing using big data analysis according to claim 1 is characterized in that: Collect the time and error rate data of chip mass production test, use big data analysis to sort and cluster the data, identify key distributions and abnormal points in the data, and obtain the time distribution analysis results in the following steps: S101: Collect time data and error rate data in chip mass production testing, organize the data, remove incomplete and damaged records, screen the data and exclude outliers, and obtain a processed data set; S102: sorting the time and error rate data of chip mass production test according to the processed data set, using time series sorting to arrange the data from early to late according to the test date, and performing secondary sorting according to the error rate, analyzing the distribution and trend of the data on the time axis, and obtaining the sorted data distribution result; S103: Based on the sorted data distribution result, the K-means clustering algorithm is used to divide the data into multiple clusters, calculate the distance from the data point to the cluster center, identify the key distribution and abnormal points in the cluster, and obtain the time distribution analysis result.

4. The vertical quantitative analysis method for chip mass production testing using big data analysis according to claim 3 is characterized in that: The formula for calculating the distance from the data point to the cluster center is: Among them, A i represents the distance from the i-th data point to the cluster center, x ij represents the value of the i-th data point in the j-th dimension, μ j represents the mean of the cluster center in the jth dimension, σ j represents the standard deviation of the jth dimension, w j represents the weight coefficient of the jth dimension, and k is the number of dimensions.

5. The vertical quantitative analysis method for chip mass production testing using big data analysis according to claim 1 is characterized in that: The time distribution analysis results are used to calculate the chip mass production test parameters in the future time period through a linear regression model, and the real-time data is compared to predict the change trend of the chip mass production parameters. The steps of obtaining the parameter change prediction results are as follows: S201: using the time distribution analysis result, selecting variables and data sets for model construction, setting model parameters, including slope and intercept, and using real-time data for model training to construct a linear regression model; S202: Based on the linear regression model, calculate and predict the chip mass production test parameters in the future time period, evaluate the prediction results using the data inferred by the linear regression model, verify the prediction ability of the linear regression model, and obtain the prediction calculation results; S203: Based on the prediction calculation results, the predicted chip mass production parameters are compared and analyzed with the real-time data to evaluate the differences and trends between the two, predict the change trends of the chip mass production parameters in the future time period, and obtain parameter change prediction results.

6. The vertical quantitative analysis method for chip mass production testing using big data analysis according to claim 1 is characterized in that: Based on the parameter change prediction results, the time and error rate data of chip mass production test are segmented by big data analysis, the volatility within multiple segments is vertically quantified and analyzed, and the data points exceeding the preset threshold are marked. The specific steps of obtaining the abnormal detection results are as follows: S301: using the parameter change prediction result and big data analysis technology, dividing the time and error rate data of chip mass production test into multiple time periods, analyzing the volatility and trend of the data within the target time range, and obtaining data segmentation results; S302: Based on the data segmentation results, vertical quantitative analysis is performed on the data within the time period, and statistical indicators of the data within multiple segments are calculated, including mean value, standard deviation and coefficient of variation, to identify key features of data volatility and obtain volatility analysis results; S303: Using the volatility analysis result, set a preset threshold, mark the data points that exceed the preset threshold, analyze potential production problems and chip mass production test anomalies, and obtain anomaly detection results.

7. The vertical quantitative analysis method for chip mass production testing using big data analysis according to claim 1 is characterized in that: Based on the abnormal detection results, vertical quantitative analysis of the error rate factors of chip mass production testing is performed, the abnormal data is compared with the preset failure mode, and the key factors causing chip failure are identified. The specific steps for obtaining the failure factor analysis results are as follows: S401: Based on the abnormal detection result, screen and extract the identified abnormal data of chip mass production test, separate the data points beyond the normal range, and generate abnormal data screening results; S402: Based on the abnormal data screening result, the screened abnormal data points are compared with the preset failure modes, the correlation between the data points and the failure modes is checked, and the comparative analysis results are obtained; S403: Using the comparative analysis results, identify key error rate factors behind the abnormal data points, analyze the impact of the key error rate factors on chip failure, and obtain failure factor analysis results.

8. The vertical quantitative analysis method for chip mass production testing using big data analysis according to claim 1 is characterized in that: The failure factor analysis results are used to quantitatively score the risks of chip mass production test batches, analyze the risk differences between batches, and conduct risk response. The specific steps for obtaining risk assessment results are as follows: S501: Based on the failure factor analysis results, select key failure factors as risk assessment indicators, assign risk scores to chips in the test batches, analyze risk levels of multiple batches, and obtain risk score records; S502: using the risk scoring records, analyzing the risk score differences between differentiated batches, identifying risky batches, risk sources and risk distribution, and obtaining batch risk analysis results; S503: Based on the batch risk analysis results, formulate targeted risk response measures, including adjusting production parameters, strengthening quality control and preventive maintenance, avoiding the probability of risk occurrence in future time periods, and obtaining risk assessment results.

9. The vertical quantitative analysis method for chip mass production testing using big data analysis according to claim 1, characterized in that: Based on the risk assessment results, the chip mass production test process and parameter settings are optimized, and the test efficiency is optimized by dynamically adjusting the test frequency and resource allocation. The specific steps of obtaining the vertical quantitative analysis results are as follows: S601: Based on the risk assessment result, analyze the risk factors in the batch, adjust the key parameter settings in the chip mass production test process according to the risk factors, including temperature, pressure and voltage, and obtain parameter adjustment results; S602: using the parameter adjustment result, adjusting the chip mass production test cycle and frequency according to the differentiated risk levels, calculating the optimized test cycle, optimizing the use efficiency of test resources, and obtaining a test frequency adjustment record; S603: Utilize the test frequency adjustment record to optimize the efficiency of the chip mass production test process, evaluate the performance of the optimized test process in reducing risks and optimizing chip yield, and obtain vertical quantitative analysis results.

10. The vertical quantitative analysis method for chip mass production testing using big data analysis according to claim 9, characterized in that: The formula for calculating the optimized test cycle is: Among them, f(x) is the optimized test cycle, R represents the risk level, D represents the number of chips to be tested, P represents the number of available test platforms, T represents the average test time, and a, b and c are weight parameters.

Images