Defect trend determination method, system and terminal based on historical data

Abstract

This invention provides a method, system, and terminal for defect trend determination based on historical data. It periodically acquires defect data after wafer scanning, uses mathematical calculations to determine whether recent defect data from the machine meets anomaly judgment rules, and obtains trend judgment results. When anomaly judgment rules are met, an alarm mechanism is triggered, sending the judgment results to the relevant responsible person for improvement planning. Finally, the trend judgment results are visualized as an anomaly list, facilitating subsequent statistical analysis of anomalies on each machine. This invention utilizes a program to quantify anomaly rules and perform periodic automatic judgment, significantly improving the accuracy and efficiency of anomaly judgment, saving manpower, and reducing the time spent on machine defect anomalies.

Description

Technical Field

[0001] This invention relates to the field of wafer defect detection technology, and in particular to a method, system and terminal for determining defect trends based on historical data. Background Technology

[0002] In existing technologies, after each wafer is scanned and inspected by a defect inspection machine, the APC (Advanced Process Control) system generates a defect map based on the defect coordinates. Technicians then randomly photograph several defects under a microscope, and engineers further determine the type and quantity of defects. If a critical defect type is found to be abnormal on a certain process machine, and the statistical results of the abnormal data show a significantly higher defect density, the concentration of defects in batches at each machine is used to assess whether the latest batches deviate from the historical baseline's normal range. However, the current practice of relying on manual judgment of all defect trends daily is not only time-consuming and subjective, but also carries the risk of errors due to human factors. This not only prolongs the response time to abnormal events but also affects production efficiency and product quality. Furthermore, the repetitive and tedious nature of the work greatly increases the likelihood of missed or incorrect judgments. Summary of the Invention

[0003] In view of the shortcomings of the prior art described above, the purpose of this invention is to provide a defect trend determination method, system and terminal based on historical data, to solve the technical problems of low efficiency, long and slow response time and misjudgment and missed judgment caused by human subjective factors in the prior art.

[0004] To achieve the above and other related objectives, this invention provides a defect trend determination method based on historical data. The method includes: periodically collecting software defect data in a wafer manufacturing environment; preprocessing the data of defect scenarios retrieved based on the software defect data to obtain a time series of the preprocessed target defect scenarios; dividing the preprocessed time series into multiple sub-sequences based on a sliding window; calculating relevant thresholds based on the preprocessed time series and setting various anomaly determination rules; determining whether each sub-sequence meets any anomaly determination rule, and triggering an early warning mechanism when it does, and obtaining the corresponding trend determination result; visualizing the trend determination result and updating the visualization chart in real time.

[0005] In one embodiment of the present invention, one of the following tools can be used to collect software defect data and retrieve defect data: a website data collection tool, a structured data collection tool, and a raw tabular data collection tool.

[0006] In one embodiment of the present invention, the step of preprocessing the defect data retrieved based on the software defect data to obtain a preprocessed time series of the target defect scenario includes: retrieving corresponding defect data based on the collected software defect data; wherein, the software defect data includes: defect report data, repair record data, and regression test result data; extracting data of the target defect scenario from the retrieved defect data and forming a time series composed of data points arranged by time, and performing data cleaning on the data points in the time series to obtain a preprocessed time series; wherein, each data point is the defect rate or defect quantity corresponding to the defect time under the target defect scenario.

[0007] In one embodiment of the present invention, the step of dividing the preprocessed time series into multiple subsequences based on a sliding window includes: performing data stabilization processing on the preprocessed time series; and using a sliding window of a fixed length set in conjunction with the target defect scenario to traverse all data points in the data stabilization time series to obtain multiple subsequences.

[0008] In one embodiment of the present invention, the step of calculating relevant thresholds based on the preprocessed time series and setting various anomaly judgment rules includes: calculating key statistical parameter values ​​of the preprocessed time series; wherein, the key statistical parameter values ​​include: P99, mean, median and standard deviation; setting relevant thresholds using the key statistical parameter values, and setting various anomaly judgment rules based on the relevant thresholds.

[0009] In one embodiment of the present invention, each anomaly determination rule includes: a moving average rule, including: a first set number of consecutively increasing data points in the subsequence; a single-point anomaly determination rule, including: a single data point greater than P99 and satisfying any other determination rule; a special determination rule, including: data points involving critical sites or critical defect types do not meet the density condition; a period determination rule, including: when the sum of the average and standard deviation of the preprocessed time series is greater than a preset value, it is determined that most data points in the subsequence within a set data point period are greater than the sum of the average and standard deviation of the preprocessed time series; and a fluctuation determination rule, including... When the average value of the preprocessed time series is not less than 1, a second set number of data points in the subsequence are greater than the average value of the time series; when the average value of the preprocessed time series is less than 1, a third set number of data points in the subsequence less than the second set number are greater than the average value of the time series; the continuous determination rule includes: there is a fourth set number of consecutively increasing data points in the subsequence and the last data point is greater than the sum of the average value and the standard deviation of the preprocessed time series; wherein, the set data point period, the first set number, the second set number, the third set number, and the fourth set number are related to the number of data points in the subsequence.

[0010] In one embodiment of the present invention, the step of visualizing the trend determination result and updating the visualization chart in real time includes: obtaining the trend determination result; the trend determination result includes: a preprocessed time series, subsequence anomaly judgment results, and anomaly judgment rule information; using a visualization tool to create a graph based on the preprocessed time series using a selected icon type, and marking the anomaly positions with the subsequence anomaly judgment results and the anomaly judgment rule information to form a visualization chart and display it, and updating the visualization chart in real time based on the new input data.

[0011] In one embodiment of the present invention, the method further includes: extracting key features from the time series after data stabilization processing; wherein the key features include: frequency indicators, periodic patterns, and product relevance; combining the key features into a feature set and displaying it visually.

[0012] To achieve the above and other related objectives, this invention provides a defect trend determination system based on historical data. The system includes: a data collection module for periodically collecting software defect data from a wafer fabrication environment; a data preprocessing module connected to the data collection module for preprocessing the defect data retrieved based on the software defect data to obtain a time series of the preprocessed target defect scenario; a window sliding module connected to the data preprocessing module for dividing the preprocessed time series into multiple sub-sequences based on a sliding window; an anomaly determination rule setting module connected to the window sliding module for calculating relevant thresholds based on the preprocessed time series and setting various anomaly determination rules; an anomaly determination and early warning module connected to the anomaly determination rule setting module for determining whether each sub-sequence satisfies any anomaly determination rule, triggering an early warning mechanism when satisfied, and obtaining the corresponding trend determination result; and a visualization module connected to the anomaly determination and early warning module for visually displaying the trend determination result and updating the visualization chart in real time.

[0013] To achieve the above and other related objectives, the present invention provides an electronic terminal, comprising: one or more memories and one or more processors; the one or more memories are used to store a computer program; the one or more processors are connected to the memories and are used to run the computer program to execute the defect trend determination method based on historical data.

[0014] As described above, this invention is a defect trend determination method, system, and terminal based on historical data, which has the following beneficial effects: This invention acquires defect data after wafer scanning is completed at regular intervals, uses mathematical calculations to determine whether recent defect data of the machine meets anomaly determination rules, and obtains trend determination results. When anomaly determination rules are met, an alarm mechanism is triggered to send the determination results to the relevant responsible person for improvement planning. Finally, the trend determination results are visualized and displayed as an anomaly list, facilitating subsequent statistical analysis of anomalies in each machine. This invention utilizes a program to quantify anomaly rules and perform timed automatic determination, greatly improving the accuracy and efficiency of anomaly determination, saving manpower, and reducing the time spent on machine defect anomalies. Attached Figure Description

[0015] Figure 1 The diagram shown is a flowchart illustrating a defect trend determination method based on historical data according to an embodiment of the present invention.

[0016] Figure 2 The diagram shown is a structural schematic of a defect trend determination system based on historical data according to an embodiment of the present invention.

[0017] Figure 3 The diagram shown is a structural schematic of a defect trend determination system platform based on historical data according to an embodiment of the present invention.

[0018] Figure 4 The diagram shown is a structural schematic of an electronic terminal according to an embodiment of the present invention. Detailed Implementation

[0019] The following specific examples illustrate the implementation of the present invention. Those skilled in the art can easily understand other advantages and effects of the present invention from the content disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and various details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that, unless otherwise specified, the following embodiments and features described therein can be combined with each other.

[0020] It should be noted that in the following description, reference is made to the accompanying drawings, which illustrate several embodiments of the invention. It should be understood that other embodiments may also be used, and changes in mechanical composition, structure, electrical system, and operation may be made without departing from the spirit and scope of the invention. The following detailed description should not be considered limiting, and the scope of the embodiments of the invention is defined only by the claims of the published patents. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the invention. Spatially related terms, such as “upper,” “lower,” “left,” “right,” “below,” “below,” “lower part,” “above,” “upper part,” etc., may be used herein to illustrate the relationship between one element or feature shown in the figures and another element or feature.

[0021] Throughout this specification, when it is said that a part is "connected" to another part, this includes not only "direct connection" but also "indirect connection" by placing other elements in between. Furthermore, when it is said that a part "includes" a certain constituent element, unless otherwise stated otherwise, this does not exclude other constituent elements, but rather means that other constituent elements may also be included.

[0022] The terms "first," "second," and "third," etc., used herein are for the purpose of describing various parts, components, regions, layers, and / or segments, but are not limiting. These terms are used only to distinguish one part, component, region, layer, or segment from others. Therefore, the "first part," "component," "region," "layer," or "segment" described below may refer to a "second part," "component," "region," "layer," or "segment" without departing from the scope of this invention.

[0023] Furthermore, as used herein, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context indicates otherwise. It should be further understood that the terms “comprising,” “including,” indicate the presence of the stated feature, operation, element, component, item, kind, and / or group, but do not preclude the presence, occurrence, or addition of one or more other features, operations, elements, components, items, kinds, and / or groups. The terms “or” and “and / or” as used herein are interpreted as inclusive, or mean any one or any combination thereof. Thus, “A, B, or C” or “A, B, and / or C” means “any one of: A; B; C; A and B; A and C; B and C; A, B, and C.” Exceptions to this definition arise only when combinations of elements, functions, or operations are inherently mutually exclusive in some manner.

[0024] This invention provides a defect trend determination method based on historical data. It periodically acquires defect data after wafer scanning and uses mathematical calculations to determine whether recent defect data from the machine meets anomaly criteria, obtaining a trend determination result. When the anomaly criteria are met, an alarm mechanism is triggered, sending the result to the responsible party for improvement planning. Finally, the trend determination result is visualized as an anomaly list, facilitating subsequent statistical analysis of anomalies on each machine. This invention utilizes a program to quantify anomaly rules and perform periodic automatic judgment, significantly improving the accuracy and efficiency of anomaly detection, saving manpower, and reducing the time spent on machine defect anomalies.

[0025] The present invention will now be described in detail with reference to the accompanying drawings, so that those skilled in the art can readily implement it. The present invention can be embodied in many different forms and is not limited to the embodiments described herein.

[0026] like Figure 1 This is a flowchart illustrating a defect trend determination method based on historical data in an embodiment of the present invention.

[0027] The method includes:

[0028] Step S1: Periodically collect software defect data in the wafer manufacturing environment.

[0029] In detail, a scheduled task is set up in the wafer fabrication environment, for example daily, to collect software defect data from the wafer fabrication environment. This software defect data comes from various defect scanning machines and includes: defect reports, repair records, and regression test results. Defect reports include defect information such as the time of the defect anomaly, machine information, and defect type. Repair records document the software defect repair process, including the repair time, the person responsible for the repair, the repair method, and the verification results after the repair. Regression test results include regression test results obtained after the software defect is repaired, to verify the repair effect. For example, a defect report corresponds to the defective lot information and a unique ID corresponding to the wafer production batch, used to identify and track the source of the defect data. This ID allows identification of which batch of data is abnormal, which can then be used for subsequent retrieval.

[0030] In one embodiment, defect data can be acquired periodically using one of a website data acquisition tool, a structured data acquisition tool, and a raw tabular data acquisition tool.

[0031] Specifically, if the software defect data comes from a website, the defect data can be collected periodically using website data collection tools; software defect data can also be collected periodically from the website using web crawlers (such as BeautifulSoup or Scrapy); or, in particular, Python scripts can be written to automatically scrape data in HTML or JSON format and extract the required fields.

[0032] If the software defect data is structured, it can be collected periodically using a structured data collection tool; or it can be obtained directly through API calls (such as the Requests library). For services that provide APIs, the Requests library can be used to directly obtain structured data, such as Jira, GitHub, or Bugzilla API.

[0033] If the software defect data is in raw list format, the defect data is collected periodically using a raw tabular data collection tool. Specifically, the pandas library is used to read, clean, and analyze defect lists in Excel or CSV format, including information such as date, level, and repair time.

[0034] This invention allows for the selection of appropriate data collection methods to acquire defect data based on actual needs and the characteristics of the data source, thereby meeting different analysis and management requirements.

[0035] Step S2: Perform data preprocessing on the defect data retrieved based on the software defect data to obtain the time series of the preprocessed target defect scenario.

[0036] In one embodiment, step S2 includes:

[0037] Retrieve relevant defect data based on the collected software defect data; specifically, locate the specific machine or production line station by using information such as the machine information and production line station where the defect occurred from the software defect data. Depending on the actual situation, select a suitable data acquisition tool, such as a website data acquisition tool, a structured data acquisition tool, or a raw table data acquisition tool, to obtain the corresponding defect data.

[0038] Data on the target defect scenario is extracted from the retrieved defect data. These data points are then arranged into a time series according to chronological order. Each data point in the time series is examined, and invalid or erroneous data points are removed. Invalid data points may include missing values, outliers, or illogical data points. This yields a pre-processed time series of the target defect scenario. Each data point represents the defect rate or number of defects at the corresponding defect time within the target defect scenario. The target defect scenario may involve specific production line sites, machine types, defect types, product types, etc.

[0039] Step S3: Based on a sliding window, the preprocessed time series is divided into multiple subsequences.

[0040] In one embodiment, step S3 includes:

[0041] The preprocessed time series data undergoes data stationarization; data stationarization is a key step in time series analysis, aiming to eliminate or reduce trends and seasonal fluctuations in the data, making the data more stable and facilitating subsequent analysis. If the original series is affected by trends or product correlations, it needs to be differencing, logarithmic transformation, or product correlation decomposition (such as Lorentz transformation of product correlation) to make it tend to be stationary.

[0042] Set a sliding window of a fixed length based on the target defect scenario, such as 8 batches or 15 batches.

[0043] A sliding window is used to traverse all data points in a time series that has undergone data stabilization, resulting in multiple subsequences. Specifically, the process begins by traversing the time series data. With each step, the window moves one data point forward, and after each window movement, the data within that window is extracted as a subsequence. As the window moves, it covers the entire series, generating a series of overlapping subsequences. This helps to capture short- to medium-term dependencies during the analysis.

[0044] In one embodiment, after step S3 is performed, key features such as defect frequency, periodic patterns, and product relevance can be extracted from the time series after data stabilization processing, and then the features can be visualized.

[0045] Specifically, methods for obtaining defect frequency characteristics include: calculating the average number or frequency of defects in each subsequence, which can reflect the trend of defect occurrence over a period of time. This helps to identify whether defects show trends of increase, decrease, or stabilization.

[0046] Methods for obtaining periodic pattern characteristics include: looking for fixed cyclical patterns in the data, such as monthly or quarterly changes, and calculating statistics within the rolling cycle. For example, peak defect periods can be calculated weekly, monthly, or quarterly to understand whether defects exhibit periodic changes.

[0047] Methods for obtaining lagged variable characteristics include introducing lag features, i.e., data points preceding the current data point, such as first-order lag (data points from the previous time step) and multiple-order lag (the previous few data points). This helps identify long-term dependencies, such as the lagged effects of changes in product relevance.

[0048] After calculating the individual features mentioned above, combining these features can generate a new feature set.

[0049] Finally, the feature set is visualized, for example, by using line charts, box plots, etc., to visually view the distribution and evolution of these features, which helps to understand the internal structure and potential patterns of the data.

[0050] Step S4: Calculate relevant thresholds based on the preprocessed time series and set various anomaly judgment rules.

[0051] In one embodiment, the step of calculating relevant thresholds based on the preprocessed time series and setting various anomaly determination rules includes:

[0052] Calculate the key statistical parameters of the preprocessed time series; these parameters include: P99, mean, median, and standard deviation. Specifically, the P99 of the preprocessed time series is calculated using Python's `quantile()` function to calculate the 99th percentile of the time series data. This metric is often used to assess extreme cases of data, especially useful when dealing with outliers or understanding the upper and lower limits of a dataset. For example, if the time series data represents the failure rate of a product, P99 can help identify anomalous time points or batches with extremely high failure rates.

[0053] The mean of a preprocessed time series can be calculated by dividing the sum of all data points in the time series by the number of data points in that time series. The mean of time series data can be calculated using Python's `mean()` function. The mean reflects the central tendency of the data. In anomaly detection, the mean can serve as a benchmark to compare whether data points deviate from the normal range.

[0054] The median (Med) is the middle value when a set of data is sorted from smallest to largest. If the dataset has an odd number of elements, the median is the middle value; if the dataset has an even number of elements, the median is the average of the two middle values. Methods for calculating the median of preprocessed time series data include using the Python keyword `median()`, which returns the median of the sorted array. Unlike the mean, the median is not affected by extreme values, making it more representative of the central tendency of a distribution, especially in the presence of outliers.

[0055] The standard deviation (Std) is equal to the square root of the average of the squares of the differences between each data point and its mean. The standard deviation measures the dispersion of a data point relative to its mean. In anomaly detection, the standard deviation can be used to determine the range of fluctuation for data points, thereby setting reasonable thresholds to identify outliers.

[0056] Use the calculated P99, mean, median, and standard deviation to set relevant thresholds and define corresponding judgment rules.

[0057] Specifically, based on the calculated statistics such as P99, mean, median, and standard deviation, we can flexibly choose to use P99, mean, median, and standard deviation individually as thresholds for judging anomalies, or derive new thresholds through combinations and further calculations. These thresholds are then used to set specific anomaly judgment rules in order to accurately identify anomalies in the data.

[0058] In one embodiment, the anomaly determination rules include:

[0059] The moving average rule includes the following: If, within a given subsequence, there exists a first predetermined number (N1) of data points exhibiting an increasing trend, then the moving average rule is satisfied, and the subsequence is detected as an anomaly. The first predetermined number needs to be determined based on the number of data points in the subsequence, and is generally greater than half of the data points in the subsequence. For example, if there are 8 data points in the subsequence, and the recent trend shows an upward trend for 5 consecutive data points, then the moving average rule is satisfied.

[0060] The single-point anomaly detection rule includes: when a single data point in a subsequence exceeds the P99 threshold and also meets any other detection rule, then the single-point anomaly detection rule is met and the data is detected as an anomaly.

[0061] Special judgment rules include: for data points of specific key sites or defect types, if the number of defects or defect rate does not meet the preset density conditions (such as the number being lower than a certain threshold), then the special judgment rules are met and the data is detected as abnormal.

[0062] The periodicity determination rule includes: if the sum of the mean and standard deviation of the time series exceeds a preset value, then further check the consecutive data points within a set data point period (e.g., five data points) in the subsequence. The set data point period needs to be determined based on the number of data points in the subsequence, generally including more than 50% of the consecutive data points. If most of the data points (e.g., more than 50% of the data points in the set data point period) are greater than the sum of the mean and standard deviation of the time series, then the periodicity determination rule is met, and it is detected as an anomaly. For example, when the mean (mean + 1 * std) >= 4, and any 4 out of 5 consecutive points out of the last 8 points are greater than mean + 1 * std.

[0063] Fluctuation determination rule, including: when the average value of the time series is not less than 1, if there are a second set number (set as N2) of data points in the subsequence that are greater than the average value of the time series, then the fluctuation determination rule is satisfied and it is detected as an anomaly. When the average value of the time series is less than 1, if the number of data points in the subsequence that is less than the second set quantity but greater than the third set number (set as N3, and N3 < N2) is greater than the average value of the time series, then the fluctuation determination rule is also satisfied and it is detected as an anomaly. The second set number needs to be determined according to the number of data points in the subsequence, generally exceeding 50% of the number of data points. For example, if mean >= 1, then the recent 5 points are all greater than mean; if mean < 1, then the recent 3 points are all greater than mean.

[0064] Continuous determination rule, including: in a certain subsequence, if there are a fourth set number of consecutive data points showing an increasing trend, and the last data point is greater than the sum of the average value and the standard deviation of the time series, then the continuous determination rule is satisfied and it is detected as an anomaly. The fourth set number needs to be determined according to the number of data points in the subsequence, generally exceeding 50% of the number of data points. For example, among the recent 8 points, there are 6 consecutive points increasing and the last point > mean + 1 * std.

[0065] It should be noted that the first set quantity, the second set data, and the fourth set data can be the same or different.

[0066] Step S5: Determine whether each subsequence satisfies any anomaly determination rule respectively, and when it is satisfied, trigger the warning mechanism and obtain the corresponding trend determination result.

[0067] In one embodiment, step S5 includes:

[0068] Determine whether each subsequence satisfies any anomaly determination rule respectively. If it is satisfied, then determine that the subsequence is an anomaly, generate an anomaly determination result, and form a trend determination result of the time series, specifically including: the preprocessed time series, the anomaly determination result, and the determination rule information on which it is based; store the trend determination result in the database for subsequent data analysis and visualization. Once an anomaly is detected, immediately trigger the warning mechanism; after triggering the warning mechanism, further trigger the email system to send an anomaly notification email to relevant personnel. The email content should include detailed information about the anomaly (such as the anomaly determination result, the determination rule information on which it is based, and the occurrence time, etc.). After receiving the email, relevant personnel conduct in-depth investigations or take corrective measures. Reply with the relevant reasons and improvement actions to facilitate the continuous tracking of anomaly defects.

[0069] In one embodiment, the visual display of the trend judgment result involves multiple links such as data collection, chart selection, graph creation, marker addition, color coding, interaction design, and regular update:

[0070] The trend determination results are obtained from the database; specifically, preprocessed time series data, including defect rate and occurrence time, are extracted from the database to obtain the anomaly determination results for each data point and the determination rules on which they are based.

[0071] Choose the appropriate chart type based on the characteristics of the data. For changes in defect rate or number of defects over time, line charts or area charts are common choices because they clearly show the trend. Depending on specific needs, other chart types, such as bar charts and pie charts, can also be considered to display data from different dimensions.

[0072] Create graphs based on preprocessed time series data using data visualization tools such as Matplotlib, Tableau, Power BI, or FineReport. Plot the preprocessed time series on a selected chart, with the X-axis representing time and the Y-axis representing the number or percentage of defects.

[0073] Based on the anomaly detection results of the subsequences and the anomaly detection rules used, add markers or annotations to key locations on the chart (such as the moments when thresholds are reached or exceeded) to highlight important turning points. Arrows, circles, and text colors can be used to improve the readability of the chart.

[0074] Use different colors or styles to distinguish normal fluctuations from problem areas, and provide clear titles and axis labels for each period or state so that users can understand the content of the chart.

[0075] To meet user interaction needs, hover prompts and click events can be provided as needed. Users can hover over or click on data points on the chart to view detailed defect information or trend analysis results.

[0076] As new data is acquired, keep the charts updated in real time to reflect the latest trend dynamics.

[0077] It should be noted that the steps in the defect trend determination method based on historical data can be implemented through trained statistical analysis models, including but not limited to time series analysis and machine learning algorithms (such as regression and random forests). In addition, this method can also include steps using Internet of Things (IoT) devices and data analysis to continuously monitor key indicators in the production process and provide real-time alerts for anomalies. Furthermore, Fault Tree Analysis (FTA), a system safety engineering method, can be used to predict potential problems in advance by analyzing the logical relationships between various components. Maintenance can be scheduled in advance based on Condition Monitoring (CM) and Remaining Life Prediction (RLP) to avoid significant losses due to minor issues. The choice of which method to use depends on the specific application scenario, available resources, and technological maturity.

[0078] Similar to the above embodiments, the present invention provides a defect trend determination system based on historical data.

[0079] The following specific embodiments are provided in conjunction with the accompanying drawings:

[0080] like Figure 2 This invention presents a schematic diagram of a defect trend determination system based on historical data, as shown in an embodiment of the present invention.

[0081] The system includes:

[0082] Data collection module 1 is used to periodically collect software defect data in the wafer manufacturing environment;

[0083] Data preprocessing module 2, connected to the data collection module 1, is used to preprocess the defect data retrieved based on the software defect data to obtain the time series of the preprocessed target defect scenario;

[0084] The sliding window module 3 is connected to the data preprocessing module 2 and is used to divide the preprocessed time series into multiple subsequences based on the sliding window.

[0085] The anomaly judgment rule setting module 4 is connected to the window sliding module 3 and is used to calculate relevant thresholds based on the preprocessed time series and set various anomaly judgment rules.

[0086] The anomaly detection and early warning module 5 is connected to the anomaly detection rule setting module 4. It is used to determine whether each subsequence meets any anomaly detection rule, and to trigger an early warning mechanism when it does, and to obtain the corresponding trend detection result.

[0087] The visualization module 6 is connected to the anomaly detection and early warning module 5, and is used to visualize the trend detection results and update the visualization charts in real time.

[0088] Since the implementation principle of this defect trend determination method based on historical data has been described in the foregoing embodiments, it will not be repeated here.

[0089] In one embodiment, the defect trend determination system based on historical data can be represented by a platform, such as... Figure 3As shown, the platform includes: a big data platform layer, which mainly implements the functions of data collection module 1 and data preprocessing module 2, that is, collecting software defect data and retrieving defect data through one of the following tools: website data collection tool, structured data collection tool, and raw tabular data collection tool, and performing data preprocessing. A trend analysis and anomaly rule setting layer, which mainly implements the functions of window sliding module 3 and anomaly judgment rule setting module 4, and also implements key feature extraction for trend analysis; finally, it also has an anomaly detection, visualization, and early warning layer; which can implement the functions of anomaly judgment and early warning modules and visualization modules.

[0090] The defect trend determination method based on historical data provided in this invention can be implemented on the terminal side or the server side. Regarding the hardware structure of the electronic terminal, please refer to... Figure 4 This is a schematic diagram of an optional hardware structure of an electronic terminal 1000 provided in an embodiment of the present invention. The terminal 1000 can be a mobile phone, computer device, tablet device, personal digital processing device, factory back-end processing device, etc. The terminal 1000 includes: at least one processor 1001, a memory 1002, at least one network interface 10010, and a user interface 1009. The various components in the device are coupled together through a bus system 1005. It is understood that the bus system 1005 is used to realize the connection and communication between these components. In addition to a data bus, the bus system 1005 also includes a power bus, a control bus, and a status signal bus. However, for clarity, in... Figure 4 The general will label all buses as bus systems.

[0091] The user interface 1009 may include a monitor, keyboard, mouse, trackball, clicker, button, touchpad, or touch screen.

[0092] It is understood that memory 1002 can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM) or programmable read-only memory (PROM), which serves as an external cache. By way of example, but not limitation, many forms of RAM are available, such as static random access memory (SRAM) and synchronous static random access memory (SSRAM). The memories described in the embodiments of this invention are intended to include, but are not limited to, these and any other suitable categories of memory.

[0093] In this embodiment of the invention, the memory 1002 is used to store various types of data to support the operation of the terminal 1000. Examples of this data include: any executable program for operation on the terminal 1000, such as the operating system 10021 and application program 10022; the operating system 10021 contains various system programs, such as the framework layer, core library layer, driver layer, etc., for implementing various basic services and handling hardware-based tasks. The application program 10022 may contain various applications, such as a media player, browser, etc., for implementing various application services. The defect trend determination method based on historical data provided in this embodiment of the invention can be included in the application program 10022.

[0094] The methods disclosed in the above embodiments of the present invention can be applied to or implemented by the processor 1001. The processor 1001 may be an integrated circuit chip with signal processing capabilities. In the implementation process, each step of the above method can be completed by the integrated logic circuit of the hardware in the processor 1001 or by instructions in the form of software. The processor 1001 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The processor 1001 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of the present invention. The general-purpose processor 1001 may be a microprocessor or any conventional processor, etc. The steps of the accessory optimization method provided in the embodiments of the present invention can be directly reflected as being executed by a hardware decoding processor, or being executed by a combination of hardware and software modules in the decoding processor. The software module may be located in a storage medium, which is located in a memory. The processor reads the information in the memory and combines it with its hardware to complete the steps of the aforementioned method.

[0095] In an exemplary embodiment, the terminal 1000 may be used to execute the aforementioned method by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), or complex programmable logic devices (CPLDs).

[0096] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented using computer program-related hardware. The aforementioned computer program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0097] In the embodiments provided in this application, the computer-readable and writable storage medium may include read-only memory, random access memory, EEPROM, CD-ROM or other optical disc storage devices, disk storage devices or other magnetic storage devices, flash memory, USB flash drive, portable hard drive, or any other medium capable of storing desired program code in the form of instructions or data structures and accessible by a computer. Additionally, any connection may be appropriately referred to as a computer-readable medium. For example, if instructions are transmitted from a website, server, or other remote source using coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of the medium. However, it should be understood that computer-readable and writable storage media and data storage media do not include connections, carrier waves, signals, or other transient media, but are intended for non-transient, tangible storage media. The disks and optical discs used in the application include compact discs (CDs), laser discs, optical discs, digital multifunction discs (DVDs), floppy disks, and Blu-ray discs, where disks typically copy data magnetically, while optical discs use lasers to copy data optically.

[0098] This invention has the following advantages:

[0099] 1. This invention quantifies abnormal rules, sets up an automatic looping program, and captures defect scanning result files in real time to achieve the purpose of querying current and historical scanning results in real time. This enables the system to automatically process in real time, reducing false positives and false negatives while reducing the workload of personnel. This can reduce the reaction time of such abnormal events and reduce the impact on wafers.

[0100] 2. Database storage of scan results addresses the timeliness and storage limitations of manual feedback.

[0101] 3. The web system provides real-time and historical query functions. Users can obtain the corresponding scan results by entering query conditions (such as time range, defect type, etc.).

[0102] 4. Timely and effectively determine abnormal trends in defect data to help analyze specific defects or machine-related defects.

[0103] In summary, the defect trend determination method, system, and terminal based on historical data of the present invention acquire defect data after wafer scanning is completed at regular intervals, and use mathematical calculations to determine whether the recent defect data of the machine meets the anomaly determination rules, thereby obtaining the trend determination result. When the anomaly determination rules are met, an alarm mechanism is triggered to send the determination result to the relevant responsible person for the formulation of an improvement plan. Finally, the trend determination result is visualized and displayed as an anomaly list, facilitating subsequent statistical analysis of the anomalies of each machine. The present invention utilizes a program to quantify anomaly rules and perform timed automatic determination, which greatly improves the accuracy and efficiency of anomaly determination, saves manpower, and reduces the time of machine defect anomalies. Therefore, the present invention effectively overcomes the various shortcomings of the prior art and has high industrial application value.

[0104] The above embodiments are merely illustrative of the principles and effects of the present invention and are not intended to limit the invention. Any person skilled in the art can modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by those skilled in the art without departing from the spirit and technical concept disclosed in the present invention should still be covered by the claims of the present invention.

Claims

1. A method for determining defect trends based on historical data, characterized in that, The method includes: Regularly collect software defect data in the wafer fabrication environment; The defect data retrieved based on the software defect data is preprocessed to obtain a time series of the preprocessed target defect scenario. Based on a sliding window, the preprocessed time series is divided into multiple subsequences; Calculate relevant thresholds based on the preprocessed time series and set various anomaly judgment rules; Each subsequence is determined to satisfy any anomaly detection rule, and an early warning mechanism is triggered when the rule is satisfied, and the corresponding trend determination result is obtained. The trend determination results are visualized and the visualization charts are updated in real time.

2. The defect trend determination method based on historical data as described in claim 1, characterized in that, Software defect data can be collected and retrieved using one of the following tools: website data collection tools, structured data collection tools, or raw tabular data collection tools.

3. The defect trend determination method based on historical data as described in claim 1, characterized in that, The step of preprocessing the defect data retrieved based on the software defect data to obtain the time series of the preprocessed target defect scenario includes: Retrieve the corresponding defect data based on the collected software defect data; wherein, the software defect data includes: defect report data, repair record data, and regression test result data; Data on the target defect scenario is extracted from the retrieved defect data and formed into a time series consisting of data points arranged by time. The data points in the time series are then cleaned to obtain a preprocessed time series. Each data point represents the defect rate or number of defects at the corresponding defect time in the target defect scenario.

4. The defect trend determination method based on historical data according to claim 3, characterized in that, The method of dividing the preprocessed time series into multiple subsequences based on a sliding window includes: Perform data stabilization on the preprocessed time series; By using a fixed-length sliding window that is combined with the target defect scenario, all data points in the time series after data stabilization are traversed to obtain multiple subsequences.

5. The defect trend determination method based on historical data according to claim 4, characterized in that, The calculation of relevant thresholds based on the preprocessed time series and the setting of various anomaly judgment rules include: Calculate the key statistical parameters of the preprocessed time series; wherein the key statistical parameters include: P99, mean, median, and standard deviation; Set relevant thresholds using key statistical parameter values, and set various anomaly judgment rules based on these thresholds.

6. The defect trend determination method based on historical data as described in claim 5, characterized in that, The rules for determining anomalies include: The moving average rule includes: the number of consecutive data points in the subsequence increases by a first set number; The single-point anomaly detection rules include: there is a single data point greater than P99 and it satisfies any other detection rule; Special judgment rules include: data points involving critical sites or critical defect types do not meet density conditions; The period determination rules include: when the sum of the mean and standard deviation of the preprocessed time series is greater than a preset value, it is determined that most data points in the subsequence within a set data point period are greater than the sum of the mean and standard deviation of the preprocessed time series. The fluctuation determination rules include: when the average value of the preprocessed time series is not less than 1, a second set number of data points in the subsequence are greater than the average value of the time series; when the average value of the preprocessed time series is less than 1, a third set number of data points in the subsequence that are less than the second set number are greater than the average value of the time series. The continuous determination rules include: there are four consecutive data points in the subsequence that increase by a set number and the last data point is greater than the sum of the mean and standard deviation of the preprocessed time series; The set data point period, the first set number, the second set number, the third set number, and the fourth set number are related to the number of data points in the subsequence.

7. The defect trend determination method based on historical data as described in claim 1, characterized in that, The visualization of the trend determination results and the real-time updating of the visualization charts include: Obtain the trend determination result; the trend determination result includes: the preprocessed time series, the subsequence anomaly determination result, and the anomaly determination rule information on which it is based; Visualization tools are used to create graphs based on preprocessed time series using selected icon types. Anomaly locations are marked by the subsequence anomaly judgment results and the anomaly judgment rules, forming and displaying a visualization chart. The visualization chart is updated in real time based on new input data.

8. The defect trend determination method based on historical data according to claim 4, characterized in that, The method further includes: Key features are extracted from time series data that have undergone data stabilization processing; wherein, the key features include: frequency indicators, periodic patterns, and product relevance; The key features are combined into a feature set and then visualized.

9. A defect trend determination system based on historical data, characterized in that, The system includes: The data collection module is used to periodically collect software defect data in the wafer manufacturing environment; The data preprocessing module, connected to the data collection module, is used to preprocess the defect data retrieved based on the software defect data to obtain the time series of the preprocessed target defect scenario. A sliding window module, connected to the data preprocessing module, is used to divide the preprocessed time series into multiple sub-series based on a sliding window; An anomaly determination rule setting module, connected to the window sliding module, is used to calculate relevant thresholds based on the preprocessed time series and set various anomaly determination rules; An anomaly detection and early warning module is connected to the anomaly detection rule setting module. It is used to determine whether each subsequence meets any anomaly detection rule, and to trigger an early warning mechanism when it does, and to obtain the corresponding trend detection result. The visualization module, connected to the anomaly detection and early warning module, is used to visualize the trend detection results and update the visualization charts in real time.

10. An electronic terminal, characterized in that, include: One or more memories and one or more processors; The one or more memories are used to store computer programs; The one or more processors are connected to the memory and are used to run the computer program to perform the method as described in any one of claims 1 to 8.

Images