An overhead shuttle vehicle self-failure detection system for semiconductor manufacturing

Abstract

The application discloses a kind of overhead shuttle car self fault detection systems for semiconductor manufacturing, specifically relates to overhead shuttle car self fault detection field, including data acquisition module, mechanical vibration anomaly detection module, positioning deviation detection module, motion stability detection module and wafer carrier docking fault detection module;Data acquisition module: by multiple types of sensors real-time acquisition overhead shuttle car operation target data;Mechanical vibration anomaly detection module: based on the target data is carried out statistical characteristic analysis to vibration signal, identifies the early wear or looseness failure of mechanical component;The present application has obtained the high-precision data reflecting the real running state of equipment by real-time collection of multidimensional operation data, achieves to identify the early failure hidden danger of mechanical vibration anomaly, positioning deviation, advances fault discovery time from after-service to real-time monitoring stage, reduces the benefit of wafer transmission delay risk.

Description

Technical Field

[0001] This invention relates to the field of self-fault detection technology for elevated shuttle vehicles, and more specifically, to a self-fault detection system for elevated shuttle vehicles used in semiconductor manufacturing. Background Technology

[0002] In the semiconductor manufacturing process, the overhead shuttle is the core equipment for wafer transportation, undertaking the task of high-precision and high-speed handling of wafers in a cleanroom environment.

[0003] However, elevated shuttles may face various failure risks during long-term operation, such as mechanical wear, sensor malfunction, and deviation from the movement trajectory. Traditional fault detection relies on regular manual inspections and post-event maintenance, which cannot perceive the equipment's operating status in real time. This can easily lead to wafer transport delays, equipment downtime losses, and even wafer contamination and scrapping, seriously affecting the continuity and yield of semiconductor production lines.

[0004] Therefore, there is an urgent need for a self-fault detection system for elevated shuttles used in semiconductor manufacturing. By deploying multi-dimensional sensors on key components of the elevated shuttle, operating parameters can be collected in real time and a fault feature space can be constructed. Combined with dynamic constraints and statistical analysis algorithms, the system can achieve real-time detection and early warning of typical faults such as abnormal mechanical vibration, positioning deviation, and deterioration of motion stability, thereby enabling real-time monitoring of the elevated shuttle's operating status and accurate fault identification. Summary of the Invention

[0005] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a self-fault detection system for elevated shuttle vehicles used in semiconductor manufacturing, which solves the problems of strong lag and ambiguous fault location in the traditional detection methods mentioned in the background art through the following solutions.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a self-fault detection system for elevated shuttle vehicles used in semiconductor manufacturing, comprising a laser rangefinder, a triaxial accelerometer, an encoder, a vision recognition terminal, and a control unit, specifically including:

[0007] Data acquisition module: Acquires target data of the elevated shuttle bus in real time through multiple types of sensors;

[0008] Mechanical vibration anomaly detection module: Based on the target data, it performs statistical feature analysis on the vibration signal to identify early wear or loosening faults of mechanical parts;

[0009] Positioning deviation detection module: It determines faults by comparing the theoretical trajectory with the actual trajectory, and detects abnormal displacement control accuracy or mechanical jamming in the drive system;

[0010] Motion stability detection module: Determines motion stability through jerk analysis and identifies abnormalities in the control stability of the drive system;

[0011] Wafer carrier docking fault detection module: It uses visual positioning technology to detect the docking accuracy between the overhead shuttle and the wafer storage or processing equipment, so as to avoid wafer damage or contamination due to docking deviation.

[0012] Preferably, the target data includes: the lateral offset of the vehicle body output by the laser rangefinder at a frequency of 200Hz. The vibration acceleration signal is synchronously output through the triaxial accelerometer. , as well as The vibration acceleration signal is then subjected to a 50Hz low-pass filter to remove environmental noise interference; the motor speed is fed back in real time via the encoder. ( ) and cumulative pulse count N(t); each time the visual recognition terminal completes a wafer carrier docking, it acquires an image of the positioning pin hole and calculates the center coordinates of the pin hole in the image coordinate system, which are then converted to world coordinate system coordinates through a calibration matrix. ).

[0013] Preferably, the statistical feature analysis includes normal vibration feature modeling, real-time vibration monitoring, and fault determination; the normal vibration feature modeling involves collecting 100 sets of normal vibration data under the unloaded and uniform speed operation state of the elevated shuttle, and calculating the root mean square value of the acceleration of each axis as the reference value, specifically expressed as follows: , , ,in, , as well as These represent the root mean square reference values ​​of vibration acceleration along the x, y, and z axes, respectively, and T represents the sampling time window. , as well as These represent the x, y, and z-axis vibration acceleration signals under normal conditions, respectively; the real-time vibration monitoring is specifically represented as follows: , , ,in, , as well as These represent the root mean square values ​​of the vibration accelerations along the x, y, and z axes, calculated in real time. ( ), ( ), ( The numbers () represent the instantaneous vibration acceleration signals in the x, y, and z axes respectively during real-time vibration monitoring of the elevated shuttle vehicle. It is a time variable, corresponding to any sampling moment within the sliding window, used to iterate through all acceleration sampling data within the window. The sliding window time is indicated; the fault determination is as follows: when the root mean square value of vibration on any axis exceeds 1.5 times the reference value, it is determined that there is abnormal mechanical vibration and a first warning is issued.

[0014] Preferably, the theoretical trajectory is calculated based on the relationship between motor speed and displacement, according to a preset running path. ,in, ( ) represents the motor speed, p represents the motor lead screw pitch; the actual trajectory: calculate the actual displacement S(t) = N(t). 0.01; The comparison is used to calculate the deviation between the actual displacement and the theoretical displacement. Simultaneously, the cumulative error of lateral offset is calculated using laser ranging data. Where n represents the number of sampling points, and express Time and The lateral spacing measurement value at any given time; the fault determination: when | |>0.5mm and If the error is greater than 0.3mm, a positioning deviation fault is determined, and a second-level warning signal is issued.

[0015] Preferably, the jerk analysis involves performing differential processing on the acceleration signal to calculate the jerk. , , ,in, , as well as These represent the jerk along the x, y, and z axes, respectively. This represents the sampling time interval; and the average value of the absolute value of the jerk is used as a stationarity index. Where SM represents the motion stability index, and m represents the number of jerk sampling points. , as well as These represent the k-th acceleration sample values ​​along the x, y, and z axes, respectively. The motion stability determination is as follows: when the stability index SM exceeds twice the average value under normal operating conditions, motion stability is determined to be deteriorated, and a third warning signal is issued.

[0016] Preferably, the visual positioning technology includes visual positioning error calculation, pin hole attitude angle deviation calculation, and fault determination; the visual positioning error calculation: defines the theoretical docking position coordinates as ( ), calculate the Euclidean distance between the actual positioning coordinates and the theoretical coordinates. The pin hole attitude angle deviation is calculated by: extracting the pin hole axis direction through image edge detection, and calculating the angle deviation between the actual axis and the theoretical axis. = ,in, This represents the actual angle of the pin hole axis obtained through image fitting. Indicates the theoretical axis angle; the fault determination: when >50μm or A fourth warning signal will be issued when the temperature is greater than 0.5°.

[0017] The technical effects and advantages of this invention are as follows:

[0018] 1. This invention deploys laser rangefinders, triaxial accelerometers, encoders, and vision recognition modules on key components of an elevated shuttle vehicle to collect multi-dimensional operational data in real time, including lateral offset of the vehicle body, vibration acceleration, displacement trajectory, and wafer carrier docking coordinates. Noise interference is removed through preprocessing algorithms such as low-pass filtering and moving average, resulting in high-precision data reflecting the true operating status of the equipment. This solves the problems of poor real-time perception of equipment status and delayed fault detection in the prior art, and achieves the benefit of early identification of mechanical vibration anomalies and positioning deviations, advancing the fault detection time from post-maintenance to the real-time monitoring stage, and reducing the risk of wafer transmission delays.

[0019] 2. This invention constructs a vibration root mean square value benchmark model, a theoretical trajectory and actual trajectory deviation algorithm, and an acceleration stability index. Combined with multi-level threshold judgment logic, it performs fusion analysis on multi-source data, achieving accurate identification capabilities for typical faults such as abnormal mechanical vibration, positioning deviation, and deterioration of motion stability. This solves the problems of ambiguous fault location and difficulty in quantifying fault severity in the background technology, achieving a four-level early warning classification. It provides clear fault level guidance for maintenance personnel, shortens fault troubleshooting time, and reduces unnecessary downtime losses. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall structure of the present invention. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] A self-fault detection system for elevated shuttle vehicles used in semiconductor manufacturing includes a laser rangefinder, a triaxial accelerometer, an encoder, a vision recognition terminal, and a control unit.

[0023] The laser rangefinder is installed on both sides of the guide rail (to detect lateral offset). It samples at a high frequency of 200Hz to obtain the lateral distance between the vehicle body and the guide rail. Lateral offset reflects guide rail wear or abnormality of the vehicle body guide wheels.

[0024] The triaxial acceleration sensor is installed at the vehicle's center of gravity and collects triaxial vibration acceleration at a sampling frequency of 1000Hz. Abnormal vibration is usually related to loose mechanical parts, worn bearings, or transmission system failure.

[0025] The encoder is integrated into the drive motor shaft and records the motor speed and pulse count in real time.

[0026] The visual recognition terminal: The linear CCD camera captures pin hole images each time the wafer carrier is docked, extracts the center coordinates of the pin holes through image processing algorithms, and then converts them into world coordinate system coordinates through a calibration matrix for detecting docking positioning deviations.

[0027] The control unit enables synchronous acquisition, preprocessing, and fault algorithm calculation of multi-source data.

[0028] As attached Figure 1 The self-fault detection system for an elevated shuttle used in semiconductor manufacturing, as shown, includes:

[0029] Data acquisition module: Acquires target data of the elevated shuttle bus in real time through multiple types of sensors;

[0030] In this embodiment, it should be specifically noted that the target data includes: the lateral offset of the vehicle body output by the laser rangefinder at a frequency of 200Hz. The vibration acceleration signal is synchronously output through the triaxial accelerometer. , as well as The vibration acceleration signal is then subjected to a 50Hz low-pass filter to remove environmental noise interference; the motor speed is fed back in real time via the encoder. ( ) and cumulative pulse count N(t); each time the visual recognition terminal completes a wafer carrier docking, it acquires an image of the positioning pin hole and calculates the center coordinates of the pin hole in the image coordinate system, which are then converted to world coordinate system coordinates through a calibration matrix. ).

[0031] Mechanical vibration anomaly detection module: Based on the target data, it performs statistical feature analysis on the vibration signal to identify early wear or loosening faults of mechanical parts;

[0032] In this embodiment, it should be specifically noted that: the statistical feature analysis includes normal vibration feature modeling, real-time vibration monitoring, and fault determination; the normal vibration feature modeling involves: collecting 100 sets of normal vibration data under the unloaded and uniform speed operation state of the elevated shuttle, and calculating the root mean square value of the acceleration of each axis as the reference value, specifically expressed as: , , ,in, , as well as These represent the root mean square reference values ​​of vibration acceleration along the x, y, and z axes, respectively, and T represents the sampling time window. , as well as These represent the x, y, and z-axis vibration acceleration signals under normal conditions, respectively. The RMS value reflects the vibration energy level, and the baseline value represents the healthy vibration energy when the equipment is fault-free. The real-time vibration monitoring is specifically expressed as follows: , , ,in, , as well as These represent the root mean square values ​​of the vibration accelerations along the x, y, and z axes, calculated in real time. ( ), ( ), ( The numbers () represent the instantaneous vibration acceleration signals in the x, y, and z axes respectively during real-time vibration monitoring of the elevated shuttle vehicle. It is a time variable, corresponding to any sampling moment within the sliding window, used to iterate through all acceleration sampling data within the window. The sliding window time indicates the dynamic changes in vibration energy, avoiding misjudgment based on data from a single moment. The fault determination is as follows: when the root mean square value of vibration of any axis exceeds 1.5 times the reference value, it is determined that there is an abnormal mechanical vibration and a first warning is issued. 1.5 times the reference value corresponds to a 125% increase in vibration energy, which is usually caused by component wear leading to a decrease in contact stiffness or an increase in clearance, such as bearing ball wear or gear tooth surface damage.

[0033] Positioning deviation detection module: It determines faults by comparing the theoretical trajectory with the actual trajectory, and detects abnormal displacement control accuracy or mechanical jamming in the drive system;

[0034] In this embodiment, it should be specifically noted that: the theoretical trajectory is calculated based on the relationship between motor speed and displacement according to the preset running path. ,in, ( () represents the motor speed, and p represents the motor lead screw pitch. The motor speed is directly proportional to the lead screw rotation angle. The pitch is converted into linear displacement, reflecting the ideal running trajectory. The actual trajectory is calculated as: actual displacement S(t) = N(t). 0.01; The comparison is used to calculate the deviation between the actual displacement and the theoretical displacement. This reflects the cumulative error of the drive system, and at the same time, the cumulative error of lateral offset is calculated using laser ranging data. Where n represents the number of sampling points, and express Time and The lateral spacing measurement at each moment, summed with the absolute value of the lateral offset at adjacent moments, reflects the degree of serpentine swaying of the vehicle body along the guide rail. An abnormal increase may be caused by guide rail deformation or guide wheel wear; the fault determination: when | |>0.5mm and If the error is greater than 0.3mm, a positioning deviation fault is determined, and a second-level warning signal is issued. The positioning accuracy required for semiconductor wafer transmission is <±1mm, and 0.5mm is the warning threshold. A lateral cumulative error >0.3mm indicates that the running trajectory deviates significantly from a straight line, which may cause a collision risk.

[0035] Motion stability detection module: Determines motion stability through jerk analysis and identifies abnormalities in the control stability of the drive system;

[0036] In this embodiment, it should be specifically noted that: the acceleration analysis involves differential processing of the acceleration signal to calculate the acceleration. , , ,in, , as well as These represent the jerk along the x, y, and z axes, respectively. The jerk represents the sampling time interval. It reflects the instantaneous change in acceleration; during smooth operation, the jerk should approach zero. A larger value indicates a more severe motion shock (such as jerking during start / brake or gear meshing impact). The average absolute value of the jerk is calculated as an indicator of smoothness. Where SM represents the motion stability index, and m represents the number of jerk sampling points. , as well as These represent the k-th acceleration sampling values ​​along the x, y, and z axes, respectively. By combining the three-axis data, single-axis interference is avoided, and the overall motion smoothness is reflected. The motion stability determination is as follows: when the stability index SM exceeds twice the average value under normal operating conditions, the motion stability is determined to be deteriorated, and a third warning signal is issued. A doubling of acceleration usually corresponds to abnormal drive system control parameters (such as PID adjustment failure) or sudden changes in mechanical resistance (such as the guide rail lubricating oil drying out), which may cause the wafer carrier to shake or even fall off.

[0037] Wafer carrier docking fault detection module: It uses visual positioning technology to detect the docking accuracy between the overhead shuttle and the wafer storage or processing equipment, so as to avoid wafer damage or contamination due to docking deviation.

[0038] In this embodiment, it should be specifically noted that: the visual positioning technology includes visual positioning error calculation, pin hole attitude angle deviation calculation, and fault determination; the visual positioning error calculation: the theoretical docking position coordinates are defined as ( ), calculate the Euclidean distance between the actual positioning coordinates and the theoretical coordinates. The pin hole attitude angle deviation is calculated by: extracting the pin hole axis direction through image edge detection, and calculating the angle deviation between the actual axis and the theoretical axis. = ,in, This represents the actual angle of the pin hole axis obtained through image fitting. Indicates the theoretical axis angle; the fault determination: when >50μm or A fourth warning signal will be issued when the temperature is greater than 0.5°.

[0039] This invention deploys laser rangefinders, triaxial accelerometers, encoders, and vision recognition modules on key components of an elevated shuttle vehicle to collect multi-dimensional operational data in real time, including lateral displacement, vibration acceleration, displacement trajectory, and wafer carrier docking coordinates. Noise interference is removed through preprocessing algorithms such as low-pass filtering and moving average, resulting in high-precision data reflecting the true operating status of the equipment. This solves the problem in the prior art where "traditional manual inspections are difficult to perceive equipment status in real time, and fault detection is lagging," enabling early identification of potential faults such as abnormal mechanical vibration (e.g., bearing wear, loose gears) and positioning deviations (e.g., motor step loss, guide rail deformation). The fault detection time is moved from post-repair to real-time monitoring, reducing the risk of wafer transmission delays. By constructing a vibration root mean square (RMS) benchmark model, a theoretical trajectory vs. actual trajectory deviation algorithm, and an acceleration stability index, combined with multi-level threshold judgment logic to fuse and analyze multi-source data, the system achieves accurate identification of typical faults such as abnormal mechanical vibration, positioning deviation, and deterioration of motion stability. This solves the problems of "fuzzy fault location and inability to quantify fault severity" in the background technology, achieving a four-level early warning classification. This provides maintenance personnel with clear fault level guidance, shortens fault troubleshooting time, and reduces unnecessary downtime losses.

[0040] Secondly: The accompanying drawings of the embodiments disclosed in this invention only involve the structures involved in the embodiments disclosed in this invention. Other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of this invention can be combined with each other.

[0041] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. An overhead shuttle self-failure detection system for semiconductor manufacturing, comprising: The system comprises a laser ranging sensor, a three-axis acceleration sensor, an encoder, a visual recognition terminal and a control unit, and specifically comprises: A data acquisition module: real-time acquisition of target data of the overhead shuttle vehicle through multiple types of sensors; The target data includes: outputting the lateral offset of the vehicle body by the laser ranging sensor at a frequency of 200 Hz ; synchronously outputting the vibration acceleration signal by the three-axis acceleration sensor 、 and , and performing 50 Hz low-pass filtering on the vibration acceleration signal to remove environmental noise interference; feeding back the motor speed ( ) and the cumulative pulse number N(t) in real time by the encoder; collecting the positioning pin hole image and calculating the pin hole center coordinates in the image coordinate system every time the wafer carrier is docked by the visual recognition terminal, and converting the world coordinate system coordinates through the calibration matrix ; A mechanical vibration anomaly detection module: statistical feature analysis of vibration signals based on the target data to identify early wear or looseness failure of mechanical components; A positioning deviation detection module: fault determination by comparing the theoretical trajectory with the actual trajectory to detect displacement control precision anomalies or mechanical jamming of the drive system; The theoretical trajectory: according to the preset running path, the theoretical displacement is calculated based on the motor speed and displacement relationship Wherein, ( ) represents the motor speed, and p represents the motor screw pitch; the actual trajectory: the actual displacement S(t)=N(t) 0.01; the comparison is used for calculating the deviation of the actual displacement and the theoretical displacement At the same time, the accumulated error of the lateral offset is calculated through the laser ranging data Wherein, n represents the number of sampling points, And represents The lateral distance measurement value at the moment and The moment; the fault determination: when |>0.5mm and >0.3mm, it is judged that there is a positioning deviation fault, and a second-level early warning signal is issued; A motion stability detection module: motion stability determination through jerk analysis to identify control stability anomalies of the drive system; A wafer carrier docking fault detection module: detection of the docking precision of the overhead shuttle vehicle with the wafer storage or processing equipment through visual positioning technology to avoid wafer breakage or contamination caused by docking deviation.

2. The overhead shuttle self-failure detection system for semiconductor manufacturing of claim 1, wherein: The statistical feature analysis includes normal vibration feature modeling, real-time vibration monitoring, and fault diagnosis. The normal vibration feature modeling involves collecting 100 sets of normal vibration data under the unloaded, uniform-speed operation of the elevated shuttle, and calculating the root mean square value of the acceleration of each axis as a baseline value, specifically expressed as follows: , , ,in, , as well as These represent the root mean square reference values ​​of vibration acceleration along the x, y, and z axes, respectively, and T represents the sampling time window. , as well as These represent the x, y, and z-axis vibration acceleration signals under normal conditions, respectively; the real-time vibration monitoring is specifically represented as follows: , , ,in, , as well as These represent the root mean square values ​​of the vibration accelerations along the x, y, and z axes, calculated in real time. ( ), ( ), ( The numbers () represent the instantaneous vibration acceleration signals in the x, y, and z axes respectively during real-time vibration monitoring of the elevated shuttle vehicle. It is a time variable, corresponding to any sampling moment within the sliding window, used to iterate through all acceleration sampling data within the window. The sliding window time is indicated; the fault determination is as follows: when the root mean square value of vibration on any axis exceeds 1.5 times the reference value, it is determined that there is abnormal mechanical vibration and a first warning is issued.

3. The overhead shuttle self-failure detection system for semiconductor manufacturing of claim 1, wherein: The jerk analysis: the jerk signal is differentiated to calculate the jerk , , wherein, , and represent the x, y and z axis jerk respectively, denotes the sampling time interval; and the stationarity index is calculated as the average of the absolute value of the jerk: wherein, SM denotes the motion stationarity index, and m denotes the number of jerk samples, , and represent the kth jerk sample value of the x, y and z axis respectively; and the motion stationarity determination: when the stationarity index SM exceeds 2 times the mean value in the normal operating state, the motion stationarity is determined to be degraded, and a third early warning signal is issued.

4. The overhead shuttle self-failure detection system for semiconductor manufacturing of claim 1, wherein: The visual positioning technology comprises visual positioning error calculation, pin-hole attitude angle deviation calculation and fault determination; the visual positioning error calculation defines a theoretical docking position coordinate as (x0, y0, z0) and calculates the Euclidean distance between an actual positioning coordinate and the theoretical coordinate as d0. ​​ The pin hole posture angle deviation calculation: the pin hole axis direction is extracted by image edge detection, and the included angle deviation between the actual axis and the theoretical axis is calculated = , wherein, represents the actual angle of the pin hole axis obtained by image fitting, represents the theoretical axis angle; the fault determination: when > 50 μm or > 0.5°, a fourth early warning signal is sent.

Images