CN-122022778-A - Machine health monitoring method, equipment and storage medium

CN122022778ACN 122022778 ACN122022778 ACN 122022778ACN-122022778-A

Abstract

The present application relates to the field of semiconductor manufacturing, and in particular, to a method and apparatus for monitoring machine health, and a storage medium. The method comprises the steps of obtaining machine data and a yield detection result of the machine data, wherein the machine data comprises a plurality of data points, carrying out anomaly monitoring on the machine data based on a current threshold interval corresponding to each data point, identifying abnormal data points suspected to be abnormal, when the yield detection result is qualified, calling verification data based on target acquisition time corresponding to the abnormal data points, target process steps and process time of the abnormal data points in the process steps, judging whether to grant updating of the current threshold interval of the abnormal data points based on the verification data, and if yes, updating the current threshold interval based on parameter values of the abnormal data points. The application effectively reduces the false alarm rate caused by harmless process fluctuation, and ensures the safety and reliability of the threshold self-adaption process, thereby coordinating the contradiction between sensitivity and accuracy of abnormal monitoring in machine health.

Inventors

LI DAN
Fu Huichu
Liang shaoyang
LI JIE
GUO YUXIANG
XIN XIAOBO

Assignees

埃克斯控股(北京)有限公司

Dates

Publication Date: 20260512
Application Date: 20260413

Claims (10)

1. The method for monitoring the health of the machine is characterized by comprising the following steps: S1, acquiring machine data and a yield detection result of the machine data, wherein the machine data comprises a plurality of data points; S2, based on the current threshold interval corresponding to each data point, carrying out anomaly monitoring on the machine data, and identifying an abnormal data point suspected to be abnormal; S3, when the yield detection result is qualified, acquiring verification data based on target acquisition time and target process step corresponding to the abnormal data point and working procedure time of the abnormal data point in the process step; s4, judging whether to grant updating of the current threshold interval of the abnormal data point or not based on the verification data; S5, if yes, updating the current threshold interval based on the parameter value of the abnormal data point; the verification data comprises an abnormal rate and a threshold adjustment record of similar data points, and at least one of using time length information and machine environment information of a target machine; S31, calling the similar data points which are in a preset time difference with the working procedure time of the abnormal data points in the target process step executed in the current time or the preset time window, and acquiring the abnormal rate and threshold adjustment record of the similar data points, S32, and acquiring the using time length information and the machine environment information of the target machine at the target acquisition time; s41, when any one or more items of verification data meet the preset trigger standard, judging to grant updating and generating a grant updating label corresponding to the abnormal data point.
2. The method according to claim 1, wherein S31 comprises: s311, acquiring local threshold adjustment records and/or local anomaly rates of similar data points in the target machine as first verification data; s312, a target machine station group is constructed, wherein the target machine station group comprises a plurality of machines which are the same as the target machine station and execute the same process steps, and group threshold adjustment records and/or group abnormality rates of similar data points in the machine station group are obtained to serve as second verification data.
3. The method of claim 2, wherein the second authentication data includes third authentication data and fourth authentication data, the S312 comprising: The method comprises the steps of obtaining a plurality of preceding machine stations with longer use time than a target machine station from the target machine station group, extracting the prior threshold adjustment record and/or the prior abnormality rate of the similar data points in the preceding machine stations as third verification data according to the information of the use time length of the target machine stations at the target acquisition time, and/or, And extracting the synchronization threshold adjustment record and/or the synchronization abnormality rate of the same class of data points in the synchronization machine as fourth verification data.
4. A method according to any one of claims 1 to 3, wherein S41 comprises: If the number of threshold adjustment records corresponding to the same type of data points is greater than a preset number of threshold values, generating a grant update label corresponding to the abnormal data points, and/or, And if the anomaly rate is higher than a preset anomaly rate threshold value, generating a grant update label corresponding to the anomaly data point.
5. The method according to any one of claims 1 to 3, wherein the machine environment information includes at least one of real-time temperature, humidity, air pressure, air flow, particle concentration, chamber cleanliness level, machine maintenance and service record, process air purity and pressure, supply voltage and current stability of a chamber in which the target machine is located, and S41 includes: Acquiring a parameter value of an environment parameter associated with an abnormal data point at a target acquisition time, and/or acquiring service time length information of a target machine at the target acquisition time as fifth verification data; and if the change rate of the parameter value is greater than a preset change threshold value and/or if the time length information is greater than a preset time length threshold value, generating a grant update label corresponding to the abnormal data point.
6. The method according to claim 1, wherein S5 comprises: performing translation adjustment on the upper limit and the lower limit of the current threshold interval based on the parameter values of the abnormal data points; or recalculating the upper limit and the lower limit of the threshold interval based on the numerical distribution of the similar abnormal data points; Or inquiring a preset mapping rule according to the using time length information and/or the machine environment information of the target machine, wherein the upper limit and the lower limit of the current threshold interval; Wherein the width of the threshold interval is kept unchanged in the adjustment process of the upper limit and the lower limit of the threshold interval.
7. The method according to claim 1, wherein the method further comprises: S6, judging that the current threshold interval of the abnormal data point is not updated based on the verification data, triggering abnormal prompt on the abnormal data point, and/or, S7, re-executing abnormal monitoring on the machine data by using the updated current threshold interval, and generating a credibility prompt for the abnormal data point if the abnormal data point is identified as normal.
8. The method according to claim 7, wherein the trust hint is determined based on the number of verification data sources and/or verification data source types that generated the grant update tag; The step S7 comprises the following steps: If the grant update label is generated only based on the first verification data, a low-credibility prompt is generated, and the updated current threshold interval is pushed to be manually checked; If the grant update tag is generated based on the fourth verification data or the fifth verification data only, generating an in-process credibility prompt, and rendering and displaying the abnormal data point with an emphasized mark in a visual interface; If the grant update tag is generated based on at least two types of verification data from different sources together or based on a combination of fifth verification data and any other verification data, a high-reliability prompt is generated, and a machine aging evaluation report, a key part replacement suggestion or preventive maintenance prompt information is output.
9. A computer device, the device comprising: A memory for storing a computer program; a processor for executing the computer program and for implementing the method for monitoring machine health according to any one of claims 1 to 8 when the computer program is executed.
10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, which when executed by a processor causes the processor to implement the method of monitoring the health of a machine according to any one of claims 1 to 8.

Description

Machine health monitoring method, equipment and storage medium Technical Field The present application relates to the field of semiconductor manufacturing, and in particular, to a method and apparatus for monitoring machine health, and a storage medium. Background In the semiconductor manufacturing process, the process stability directly determines the yield and performance consistency of the chip. In order to ensure production quality, wafer factories generally need to monitor key process parameters (such as film thickness, critical dimension CD, doping concentration, electrical parameters, etc.) in real time. For example, the invention patent application with the publication number of CN121009451A discloses an abnormality identification method and system combined with hierarchical parameter analysis, and relates to the technical field of fault detection; searching in a fault record database to determine a first sub-event set, extracting key parameters to obtain a first sub-event key parameter value sequence set, carrying out parameter fluctuation duration integration analysis to obtain a first parameter fluctuation duration, extracting associated equipment to obtain K secondary fault component sets, obtaining K secondary operation parameter sets, obtaining an abnormal component set and an abnormal detection item set, and carrying out potential fault detection of target semiconductor equipment. For example, the patent application with the publication number of CN118197960A discloses a semiconductor production data monitoring method and system based on a multi-mode intelligent agent, which are applied to a data monitoring system, wherein the data monitoring system comprises a data acquisition intelligent agent, a data fusion intelligent agent, a data analysis intelligent agent and a process analysis intelligent agent; the data fusion agent receives the original production data, performs data preprocessing on the original production data to obtain production data, and sends the production data to the data analysis agent, the data analysis agent receives the production data, performs data analysis on the production data to generate a data analysis result, and sends the data analysis result to the process analysis agent, and the process analysis agent receives the data analysis result, evaluates the process of semiconductor production based on the data analysis result to generate a process evaluation result. However, as process nodes are continuously miniaturized, process windows are increasingly narrow, parameter fluctuations in the manufacturing process are increasingly sensitive, and meanwhile, non-critical changes introduced by factors such as equipment aging and environmental disturbance are significantly increased, so that difficulty in anomaly determination is significantly increased. The traditional scheme generally sets a fixed upper and lower control limit or specification threshold interval based on historical data, and when the data points collected by the machine station exceed the interval, the machine station is judged to be abnormal and an alarm is triggered, so that response measures such as line stopping, investigation or equipment maintenance are started. The scheme lacks the discrimination capability of the actual influence of the abnormality, is easy to misjudge harmless fluctuation as a fault, and has high misinformation rate, rising response cost and lower practicability. Disclosure of Invention The application mainly aims to provide a method, equipment and a storage medium for monitoring machine health. In order to solve the technical problems, the application adopts the following technical scheme: the first aspect of the present application provides a method for monitoring machine health, the method comprising: S1, acquiring machine data and a yield detection result of the machine data, wherein the machine data comprises a plurality of data points; S2, based on the current threshold interval corresponding to each data point, carrying out anomaly monitoring on the machine data, and identifying an abnormal data point suspected to be abnormal; S3, when the yield detection result is qualified, acquiring verification data based on target acquisition time and target process step corresponding to the abnormal data point and working procedure time of the abnormal data point in the process step; s4, judging whether to grant updating of the current threshold interval of the abnormal data point or not based on the verification data; S5, if yes, updating the current threshold interval based on the parameter value of the abnormal data point; the verification data comprises an abnormal rate and a threshold adjustment record of similar data points, and at least one of using time length information and machine environment information of a target machine; S31, calling the similar data points which are in a preset time difference with the working procedure time of the abnormal data points in the target