Anomaly detection device, anomaly detection method, and program

Abstract

To provide an abnormality detection device that can detect abnormal operation of a shutter device more quickly.SOLUTION: Provided is an abnormality detection device for detecting abnormal operation of a shutter device that blocks light, and the device has a detection unit for detecting an abnormal operation using an abnormality detection model that outputs determination data for detecting an abnormal operation by inputting information about measurement data of the shutter device.SELECTED DRAWING: Figure 1

Description

[Technical Field]

[0001] The present invention relates to an abnormality detection device, an abnormality detection method, and a program. [Background technology]

[0002] A shutter device can be used to block light in manufacturing apparatuses such as an exposure apparatus that exposes a substrate through an original plate, and a molding apparatus that brings a composition on a substrate into contact with a mold and cures the composition by irradiating it with light.

[0003] Patent Document 1 discloses an invention relating to an exposure apparatus that is used to project a reticle pattern onto a wafer for exposure in the semiconductor manufacturing process, and that is equipped with a shutter device for controlling the amount of exposure light required during exposure. [Prior art documents] [Patent documents]

[0004] [Patent Document 1] Japanese Patent Application Laid-Open No. 2001-44110 Summary of the Invention [Problem to be solved by the invention]

[0005] Patent Document 1 describes that a lack of current in a motor that drives a shutter device is detected, and necessary preventive measures are taken based on the detection results.

[0006] However, abnormal operation occurring in a shutter device is not limited to abnormal operation caused by a lack of current in the motor. For example, abnormal operation may occur due to components such as the shutter blades or the motor shaft. For example, shutter device components deteriorate over time, which may cause abnormalities in the shutter device components. Furthermore, for example, if the shutter blades are deformed due to external force, heat, or the like, the shutter may come into contact with other components, potentially damaging the shutter or other components. Therefore, it is desirable to detect abnormal operation of a shutter device more quickly.

[0007] SUMMARY OF THE INVENTION It is therefore an object of the present invention to provide an abnormality detection device, an abnormality detection method, and a program that can more quickly detect abnormal operation of a shutter device. [Means for solving the problem]

[0008] An aspect of the present invention for solving the above problem is an abnormality detection device that detects abnormal operation of a shutter device that blocks light, and includes a detection unit that detects abnormal operation using an abnormality detection model that receives information about measurement data of the shutter device and outputs judgment data for detecting abnormal operation. The information about the measurement data includes a feature amount calculated from the measurement data, and the feature amount includes information about the center of gravity calculated from a waveform of time-series data acquired by measuring a physical quantity related to the shutter device. . [Effects of the Invention]

[0009] According to the present invention, it is possible to provide an abnormality detection device, an abnormality detection method, and a program that can more quickly detect abnormal operation of a shutter device. [Brief explanation of the drawings]

[0010] [Figure 1] FIG. 2 is a diagram illustrating a configuration of a shutter device. [Figure 2] FIG. 10 is a diagram showing an abnormality detection device that detects abnormal operation of a shutter device. [Figure 3] 4 is a flowchart showing an abnormality detection process according to the first embodiment. [Figure 4] FIG. 10 is a diagram illustrating feature amounts and Mahalanobis distances. [Figure 5] FIG. 10 is a diagram showing time-series data measured by a sensor. [Figure 6] 10 is a flowchart showing an abnormality detection process according to the second embodiment. [Figure 7] FIG. 1 is a diagram illustrating division and grouping of time series data. [Figure 8] FIG. 10 is a diagram illustrating feature amounts and Mahalanobis distances for each group. [Figure 9] FIG. 1 is a diagram showing an exposure apparatus incorporating a shutter device. [Figure 10] FIG. 10 is a diagram showing a molding device incorporating a shutter device. DETAILED DESCRIPTION OF THE INVENTION

[0011] Hereinafter, embodiments will be described in detail with reference to the accompanying drawings. Note that the following embodiments do not limit the invention according to the claims. Although multiple features are described in the embodiments, not all of these multiple features are necessarily essential to the invention, and multiple features may be combined arbitrarily. Furthermore, in the accompanying drawings, the same reference numerals are used to designate the same or similar components, and redundant explanations will be omitted.

[0012] First Embodiment FIG. 1 is a diagram showing the configuration of a shutter device. FIG. 1(a) is a cross-sectional view of shutter device 100, and FIG. 1(b) is a plan view of shutter 105. Shutter device 100 includes shutter 105. Shutter device 100 can be configured to block or pass light 102 by opening and closing the optical path of light 102 emitted from a light source (not shown) using shutter 105. Shutter 105 can be made of a metal such as aluminum or an aluminum alloy, but may also be made of other materials.

[0013] The shutter device 100 has a shutter 105 that blocks light 102, and a drive mechanism that drives the shutter 105, such as a rotation mechanism 104 that rotates the shutter 105 attached to a shaft 103 about a rotation axis RA. The rotation mechanism 104 can be, for example, a motor that rotates the shaft 103 about the rotation axis RA. The rotation axis RA may be, for example, parallel to the direction of the chief ray of the light 102 (the direction of the optical axis of an illumination optical system not shown), or may have a certain inclination relative to the direction of the chief ray of the light 102. In the following, the direction parallel to the direction of the chief ray of the light 102 is defined as the Z-axis direction, and directions that are perpendicular to each other in a plane perpendicular to the Z-axis direction are defined as the X-axis direction and the Y-axis direction.

[0014] The shutter 105 is fixed to the shaft 103 of the rotation mechanism 104 by a fixing jig 106. For example, the fixing jig 106 has a plurality of fixing parts (not shown), such as pins and bolts, and the shutter 105 is fixed to the shaft 103 by fastening the plurality of fixing parts. In the example of FIG. 1(b), the shutter 105 is provided with three blades (light-shielding portions). In this case, the rotation mechanism 104 rotates the shutter 105 by 60 degrees from a state in which the shutter 105 blocks the light 102 as a drive to open the optical path of the light 102 (open drive). In addition, the rotation mechanism 104 rotates the shutter 105 by 60 degrees from a state in which the shutter 105 transmits the light 102 as a drive to close the optical path of the light 102 (close drive). In other words, when performing one open / close drive, the rotation mechanism 104 rotates the shutter 105 by 120 degrees. Furthermore, when performing three opening and closing drives, the rotation mechanism 104 rotates the shutter 105 through 360 degrees. Furthermore, the number of blades of the shutter 105 is not limited to three and can be any number. In this case, the rotation angle of the shutter 105 in the opening and closing drive is determined according to the number of light-blocking portions of the shutter 105. In this manner, the rotation mechanism 104 drives the shutter 105 to open and close by rotating the shutter 105. Furthermore, the rotation mechanism 104 drives the shutter 105 to open and close three times per revolution of the shutter 105. Here, when the rotation mechanism 104 repeatedly drives the shutter 105 to open and close at regular intervals, the shutter 105 is driven periodically. In this manner, the period when the shutter 105 is driven periodically is referred to as the drive period.

[0015] The shutter device 100 also has a sensor 101 for measuring physical quantities related to the shutter device 100. The sensor 101 may be, for example, a sensor installed in the shutter device 100 that measures vibrations of the shutter device 100. Specifically, the sensor 101 may be a sensor that measures at least one of the position, velocity, and acceleration of (a component of) the shutter device 100. The sensor 101 may also be a sensor that measures at least one of the position, velocity, and acceleration of the shutter device 100 in a non-contact manner, such as an interferometer or an encoder. The sensor 101 may also be, for example, a sound sensor that measures sound generated from the shutter device 100. The sensor 101 may also be a torque sensor that measures rotational torque generated by the rotation mechanism 104. The sensor 101 may also be a current sensor that measures current consumption by the rotation mechanism 104. In this way, the sensor 101 measures physical quantities related to the shutter device, and measurement data of the measured physical quantities is obtained.

[0016] Furthermore, the shutter device 100 may have a drive mechanism that drives the shutter 105 along a predetermined direction instead of a rotation mechanism that drives the shutter 105 in a rotational direction. In this case, the shutter 105 does not need to have the shape shown in FIG. 1(b), as long as it has a shape that can block the light 102 by driving it along the predetermined direction. The drive mechanism, for example, blocks and passes the light 102 by reciprocating the shutter 105 along the predetermined direction. In other words, in this case too, the drive mechanism periodically drives the shutter 105 at a predetermined drive cycle by reciprocating the shutter 105.

[0017] In this way, the shutter device 100 periodically drives the shutter 105 at a predetermined drive cycle, and therefore, if it is operated for a long period of time, the parts of the shutter device 100 may wear out or deform, which may cause a breakdown of the shutter device 100. Furthermore, it is desirable to prevent breakdowns of the shutter device 100 by performing maintenance such as replacing parts before the shutter device 100 stops due to a breakdown of the shutter device 100.

[0018] Therefore, the shutter device 100 according to this embodiment quickly detects abnormal operation of the shutter device 100, so that maintenance can be performed before the shutter device 100 breaks down.

[0019] FIG. 2 illustrates an anomaly detection device that detects abnormal operation of a shutter device. The anomaly detection device 200 is configured as a computer (information processing device). The anomaly detection device 200 includes a processing unit 201 having a processor such as a CPU that performs calculations for control according to a program. The anomaly detection device 200 also includes a storage unit 202 such as a ROM that stores a control program and fixed data, and a RAM that stores a work area for the processing unit 201 and temporary data. The anomaly detection device 200 may also include a magnetic storage device (HDD) that can store larger amounts of data than ROM and RAM. The anomaly detection device 200 may also include a drive device that loads external media such as CDs, DVDs, and memory cards and reads and writes data from and to the external media. In this embodiment, the storage unit 202 is at least one of a ROM, RAM, magnetic storage device, and drive device, and the storage unit 202 stores the control program, fixed data, a work area for the processing unit, and temporary data. The anomaly detection device 200 may also include a display device 213 such as a CRT, LCD display, or touch panel. Here, the abnormality detection device may be provided inside the shutter device 100 or outside the shutter device 100.

[0020] Next, the abnormality detection process in the processing unit 201 will be described with reference to Fig. 2 and Fig. 3. Fig. 3 is a flowchart showing the abnormality detection process according to this embodiment. Here, the abnormality detection process shown in Fig. 3 is executed by each unit in the processing unit 201.

[0021] In step 301, the processing unit 201 acquires generation data 208 for generating a model (abnormality detection model) that outputs judgment data for detecting abnormal operation. First, the processing unit 201 acquires measurement data measured by the sensor 101 while the shutter device 100 is operating normally. Here, the measurement data measured by the sensor 101 includes time-series data of physical quantities related to the shutter device 100. Then, the conversion unit 203 converts the acquired measurement data from an analog signal to a digital signal (discrete signal).

[0022] Next, the calculation unit 204 calculates feature quantities from the digital signal converted by the conversion unit 203. Here, the feature quantities are calculated as the centroid values ​​of the frequency spectrum waveforms obtained by, for example, performing a discrete Fourier transform on the time-series digital signal to acquire frequency spectrum waveforms. In this way, feature quantities can be calculated by reducing the dimensions of the measurement data as information about the measurement data measured by the sensor 101. A processing mode for model generation is selected by the selection unit 205, and the calculated feature quantities are stored in the storage unit 202 as generated data 208. Here, when the selection unit 205 selects a processing mode for anomaly detection or anomaly classification, which will be described later, the feature quantities are stored in the storage unit 202 as data different from the generated data 208.

[0023] Here, a method for calculating the center of gravity of a frequency spectrum waveform as a feature quantity will be described. x The center of gravity of the amplitude component in the frequency spectrum waveform is G y In addition, the frequency f i , the frequency spectrum signal intensity L i The sum of the signal strength of the entire frequency spectrum waveform is S, and the total number of data points of the frequency spectrum waveform is n. The center of gravity of the frequency component G x , and the center of gravity of the amplitude component G y are calculated using the following formulas (1) and (2), respectively.

[0024]

number

[0025] Furthermore, the calculated feature quantity is not limited to the center of gravity of the frequency spectrum waveform. For example, the RMS (Root Mean Square) value, the peak (maximum or minimum) value of the signal strength, or the crest factor (crest factor) value expressed as the ratio of the peak value to the RMS value may be calculated as a feature quantity for the time series data. Furthermore, the result of principal component analysis using multivariate time series data consisting of multiple time series data such as vibration, torque, and control current as input may also be used as a feature quantity.

[0026] In step 302 , the generator 209 generates an anomaly detection model using the generated data 208 stored in the storage unit 202 .

[0027] Here, when the shutter device 100 operates, the probability of abnormal operation occurring is usually very low, and the time during which abnormal operation occurs is shorter than the time during which the shutter device 100 operates normally. Therefore, it is difficult to collect the amount of data measured when abnormal operation occurs (abnormal data) required to generate an abnormality detection model. For this reason, in this embodiment, a model for detecting abnormal operation is generated based on normal data. In this embodiment, anomalies are detected from a set of normal data using the Mahalanobis-Taguchi method (MT method), a typical discrimination method that combines quality engineering and multivariate analysis.

[0028] FIG. 4 is a diagram showing feature amounts and Mahalanobis distances. The MT method is a technique for detecting abnormal data based on normal data. Input data 401 is data for determining whether it represents an abnormal operation, and a normal data set 402 is a set of data with normal feature amounts. The horizontal axis represents feature amount X, and the vertical axis represents feature amount Y, and the positions are determined according to the feature amounts of each data. In the MT method, a Mahalanobis distance 403 from a reference (e.g., the center) of the normal data set 402 to the input data 401 is calculated, and if it exceeds a predetermined allowable range 404, the input data 401 is determined to be an abnormal operation.

[0029] In the MT method, a model that outputs the Mahalanobis distance 403 as judgment data is used as an anomaly detection model. The Mahalanobis distance 403 is calculated using the covariance matrix A of the input data (x, y) that is the target for detecting abnormal behavior and the normal data set (X, Y). The covariance S of the normal data set xy is calculated by equation (3).

[0030]

number

[0031] The covariance matrix A of the input data (x, y) is calculated from the variances Sx, Sy, and Sxy of the normal data using equation (4).

[0032]

number

[0033] Here, the Mahalanobis distance 403 of the input data (x, y) to the normal data set (X, Y) is calculated by equation (5) where the Mahalanobis distance 403 is MD.

[0034]

number

[0035] In this way, in step 302, the generation unit 209 generates, as an anomaly detection model, the formula (5) for inputting the input data (x, y) and outputting the Mahalanobis distance 403. Here, in this embodiment, the input data (x, y) is used as the feature quantity (G x , G y ) is used.

[0036] In step 303, the processing unit 201 acquires detection data 207 for detecting abnormal operation. The processing unit 201 acquires time-series data of physical quantities measured by the sensor 101 while the shutter device 100 is operating, and calculates feature amounts. Note that the acquisition of time-series data and calculation of feature amounts in step 303 are similar to those in step 301, and therefore detailed explanations will be omitted. Furthermore, the selection unit 205 selects a processing mode for abnormal detection, and the calculated feature amounts are stored in the storage unit 202 as detection data 207 for detecting abnormal operation.

[0037] In step 304, the detection unit 210 determines whether an abnormal operation has occurred in the shutter device 100. Specifically, the detection unit 210 uses the abnormality detection model generated by the generation unit 209 to determine whether the detection data 207 stored in the storage unit 202 is abnormal data. The detection unit 210 inputs the detection data 207 into the abnormality detection model and obtains the Mahalanobis distance 403 output from the abnormality detection model. The threshold value representing the boundary of the allowable range 404 in FIG. 4 is used as the MD th Then, the Mahalanobis distance 403 is the threshold MD th If this is the case (if the following formula (6) is satisfied), the detected data 207 is determined to be abnormal data.

[0038]

number

[0039] Here, the method of anomaly detection is not limited to the method using the MT method, and anomaly detection may be performed using, for example, the k-nearest neighbor method, One-Class SVM, AutoEncoder, or the like.

[0040] If it is determined in step 304 that abnormal operation has occurred in the shutter device 100, the processing unit 201 proceeds to step 305, and if it is not determined that abnormal operation has occurred in the shutter device 100, the processing unit 201 returns to step 303.

[0041] In step 305, the processing unit 201 acquires time-series data of physical quantities measured by the sensor 101 while the shutter device 100 is operating, and calculates feature amounts. Here, the time-series data acquired in step 305 is divided into multiple pieces of time-series data for each drive cycle of the shutter device 100, and multiple feature amounts corresponding to each of the multiple pieces of time-series data are calculated. In other respects, the acquisition of the time-series data and the calculation of the feature amounts are the same as in step 301, so detailed explanations will be omitted. In addition, the selection unit 205 selects an abnormality classification processing mode for the multiple calculated feature amounts, and the selection unit 205 stores the calculated feature amounts as classification data 206 for classifying abnormal operations in the storage unit 202.

[0042] In step 306, the classification unit 211 calculates judgment data based on at least two feature amounts selected from the plurality of feature amounts in the classification data 206. The classification unit 211 may calculate, for example, the difference or ratio between the two feature amounts as the judgment data. Alternatively, the classification unit 211 may select, for example, a plurality of combinations of two feature amounts from the plurality of feature amounts. The classification unit 211 may then calculate a plurality of differences or ratios for each combination of the two feature amounts, and calculate a statistical value such as the maximum value, minimum value, average value, or median value as the judgment data.

[0043] Here, in this embodiment, abnormal operations that occur in the shutter device 100 are classified by focusing on the fact that the shutter device 100 is driven periodically at a predetermined drive cycle. Abnormal operations of the shutter device 100 include abnormal operations that occur when the shutter 105 is deformed due to heat or external force, for example. If the shutter 105 is deformed, the shutter 105 may come into contact with components such as the housing of the shutter device 100 when the shutter 105 is driven. For example, in the case of a shutter 105 having three blades as shown in FIG. 3(b), if one blade deforms and comes into contact with a component of the shutter device 100, one abnormal operation will occur in the time-series data measured by the sensor 101 for every three opening and closing drives (one drive cycle).

[0044] FIG. 5 is a diagram showing time-series data measured by the sensor 101. FIG. 5(a) is a diagram showing time-series data 501 in the case where an abnormal operation occurs in which one blade of the shutter 105 deforms and comes into contact with a component of the shutter device 100. In the time-series data 501, time-series data 502 to 507 are time-series data at the timings when the shutter 105 is driven to open and close. In the example of FIG. 5(a), an abnormality occurs in the time-series data 502 and 505, and the waveforms of the respective time-series data are similar, and the correlation between the feature amounts calculated from the respective time-series data is high. In other words, if an abnormal operation occurs due to the shutter 105 coming into contact with a component of the shutter device 100 while the shutter 105 is being driven, periodic features will appear in the waveform of the time-series data 501 measured by the sensor 101.

[0045] Furthermore, abnormal operation of the shutter device 100 may occur, for example, when the force fixing the shutter 105 to the shaft 103 is reduced due to a decrease in the fastening force of the fixing parts of the fixing jig 106, and the shutter 105 does not completely follow the rotation of the shaft 103. In this case, the shutter 105 is unlikely to immediately come into contact with parts of the shutter device 100 when it is driven.

[0046] 5(b) is a diagram showing time-series data 508 in a case where an abnormality occurs due to a decrease in the force fixing the shutter 105 to the shaft 103, causing the shutter 105 to not completely follow the rotation of the shaft 103. In the time-series data 508, time-series data 509 to 514 represent the timings at which the shutter 105 is opened and closed. In the example of FIG. 5(b), an abnormality occurs in the time-series data 509 to 514, but as in the example of FIG. 5(a), the waveforms of the respective time-series data are not similar, and the correlation between the feature amounts calculated from the respective time-series data is low. In other words, if an abnormal operation occurs due to a decrease in the force fixing the shutter 105 to the shaft 103 when the shutter 105 is driven, no periodic feature appears in the waveform of the time-series data 501 measured by the sensor 101.

[0047] Therefore, the difference or ratio between the feature amount acquired from the time series data 502 and the feature amount acquired from the time series data 505 becomes small, but the difference or ratio between the feature amount acquired from the time series data 509 and the feature amount acquired from the time series data 502 becomes large. In other words, by using the determination data calculated based on the multiple feature amounts in the classification data 206, it is possible to classify abnormal operations that occur in the shutter device 100.

[0048] Here, which feature to select from the multiple feature values ​​in the classification data 206 is determined based on the characteristics of the shutter device 100. For example, in the example of FIG. 1(b), since there are three blades (light-shielding portions), feature values ​​calculated from time-series data for three opening and closing drives (one rotational drive), such as time-series data 502 and 505, are selected. The example of FIG. 5(a) shows an example in which six opening and closing drives (two rotational drives) are performed and time-series data 501 is acquired. In step 306, the classification unit 211 selects a combination of time-series data 502 and 505, a combination of time-series data 503 and 506, and a combination of time-series data 504 and 507. Then, the classification unit 211 calculates judgment data based on the feature values ​​calculated from each combination. Furthermore, the classification unit 211 may calculate the judgment data as multiple pieces of data, or may calculate the judgment data as statistical values ​​such as the maximum, minimum, average, or median of the multiple pieces of data.

[0049] Returning now to the description of Fig. 3, in step 306, the classification unit 211 determines whether the judgment data is equal to or less than a predetermined threshold. If it is determined that the judgment data is equal to or less than the predetermined threshold, the classification unit 211 proceeds to step 308. If it is not determined that the judgment data is equal to or less than the predetermined threshold, the classification unit 211 proceeds to step 309.

[0050] In step 308, the classification unit 211 determines that a periodic abnormal operation has occurred in the shutter device 100. Furthermore, the classification unit 211 may cause the display unit 213, via the control unit 212, to display an error message indicating that a periodic abnormal operation has occurred. Furthermore, although an abnormal operation that occurs when the shutter 105 comes into contact with a component of the shutter device 100 has been described as an example of a periodic abnormal operation, the periodic abnormal operation is not limited to this. For example, a periodic abnormal operation may be an abnormal operation that occurs when a component driven by a component other than the shutter 105 comes into contact with another component of the shutter device 100.

[0051] Furthermore, if periodic abnormal operation occurs in the shutter device 100, contact between a driven part such as the shutter 105 will cause the abnormal operation, which is highly likely to lead to damage to the shutter device 100. For this reason, the control unit 212 may perform control to stop the operation of the shutter device 100. This allows the shutter device 100 to be maintained more quickly.

[0052] In step 309, the classification unit 211 determines that a non-periodic abnormal operation has occurred in the shutter device 100. Furthermore, the classification unit 211 may cause the display unit 213, via the control unit 212, to display an error message indicating that a non-periodic abnormal operation has occurred. Furthermore, although the non-periodic abnormal operation has been described as an abnormal operation that occurs due to a decrease in the force fixing the shutter 105 to the shaft 103, the non-periodic abnormal operation is not limited to this. The non-periodic abnormal operation may be, for example, an abnormal operation that occurs due to a decrease in the force fixing a component other than the shutter 105.

[0053] Furthermore, when a non-periodic abnormal operation occurs in the shutter device 100, it is unlikely that the abnormal operation is caused by contact with a driven part such as the shutter 105, and it is unlikely that this will immediately lead to damage to parts of the shutter device 100. Therefore, the control unit 212 may perform control so that the shutter device 100 continues to operate. This makes it possible to improve the operating rate of the shutter device 100.

[0054] Furthermore, when a non-periodic abnormal operation occurs in the shutter device 100 and the control unit 212 controls the shutter device 100 to continue operating, the control unit 212 may reduce the speed of the shutter driven by the drive mechanism. Furthermore, when a non-periodic abnormal operation occurs in the shutter device 100 and the control unit 212 controls the shutter device 100 to continue operating, the control unit 212 may reduce the frequency at which the shutter is driven by the drive mechanism. This makes it possible to further improve the operating rate while reducing the possibility of damage to parts of the shutter device 100.

[0055] As described above, according to the shutter device of this embodiment, feature amounts are calculated from the measured time-series data, and an abnormality detection model is used to determine whether an abnormal operation is occurring in the shutter device, so that abnormal operations can be detected more quickly. Furthermore, judgment data is calculated based on the feature amounts calculated from the divided time-series data, and the judgment data is compared with a threshold value, so that abnormal operations occurring in the shutter device can be classified.

[0056] Second Embodiment Next, a shutter device according to a second embodiment will be described. Matters not mentioned here may follow those of the first embodiment. In this embodiment, time-series data measured by a measurement unit is divided into multiple groups, and an abnormality detection model generated for each divided group is used to determine whether an abnormal operation is occurring in the shutter device.

[0057] 6 is a flowchart showing the abnormality detection process according to this embodiment. The abnormality detection process shown in FIG.

[0058] In step 601, the processing unit 201 acquires generation data 208 for generating an abnormality detection model for detecting abnormal operation. First, the processing unit 201 acquires time series data of physical quantities measured by the sensor 101 while the shutter device 100 is operating normally. Next, the processing unit 201 divides the acquired time series data into data for each opening and closing drive of the shutter 105, and separates the divided time series data into a plurality of groups according to the order of opening and closing drive of the shutter 105 within the drive cycle of the shutter device 100.

[0059] FIG. 7 is a diagram showing division and grouping of time-series data. In the example of FIG. 7, the shutter 105 has three blades, and is opened and closed three times per cycle. The example of FIG. 7 also shows time-series data 700 for two cycles. In the first cycle T1, the time-series data measured during the first open / close drive is designated 701a, the time-series data measured during the second open / close drive is designated 702a, and the time-series data measured during the third open / close drive is designated 703a. In the second cycle T2, the time-series data measured during the first open / close drive is designated 701b, the time-series data measured during the second open / close drive is designated 702b, and the time-series data measured during the third open / close drive is designated 703b.

[0060] The processing unit 201 divides the acquired time series data 700 into time series data 701a, 702a, 703a, 701b, 702b, and 703b. Then, the processing unit 201 divides the time series data 701a and 701b into group 1, the time series data 702a and 702b into group 2, and the time series data 703a and 703b into group 3.

[0061] 7, the shutter 105 is opened and closed three times per cycle, but the present invention is not limited to this and any number of cycles may be performed. Furthermore, in the example of FIG. 7, time series data for two cycles is acquired, but the present invention is not limited to this and any number of cycles may be acquired. Furthermore, the number of groups into which the time series data is divided is not limited to three, and the time series data is divided into a number of groups according to the number of times the shutter 105 is opened and closed per cycle.

[0062] Next, the conversion unit 203 converts the grouped time series data from analog signals to digital signals, and the calculation unit 204 calculates feature amounts for each group from the digital signals converted by the conversion unit 203. The method for calculating feature amounts is the same as in the first embodiment, and therefore a detailed description thereof will be omitted.

[0063] Fig. 8 is a diagram showing the feature amounts and Mahalanobis distances for each group. Fig. 8 shows a case where the data is divided into three groups, with Fig. 8(a) showing the feature amounts for group 1, Fig. 8(b) showing the feature amounts for group 2, and Fig. 8(c) showing the feature amounts for group 3. Normal data sets 801 to 803 are sets of normal feature amount data belonging to groups 1 to 3, respectively. The calculated and grouped feature amounts are then used to select a model generation processing mode, and are stored for each group in the storage unit 202 as generated data 208.

[0064] Returning now to the description of Fig. 6, in step 602, the generation unit 209 generates an anomaly detection model for each group using the generated data 208 stored in the storage unit 202. The method for generating the anomaly detection model is the same as in the first embodiment, and therefore a detailed description thereof will be omitted.

[0065] In step 603, the processing unit 201 acquires detection data 207 for detecting abnormal operation. First, the processing unit 201 acquires time-series data of physical quantities measured by the sensor 101 while the shutter device 100 is operating. Next, the processing unit 201 divides the acquired time-series data for each opening and closing drive of the shutter 105, and separates the divided time-series data into a plurality of groups according to the timing of opening and closing the shutter 105. Note that the acquisition of time-series data and calculation of feature amounts in step 603 are similar to step 601, and therefore detailed description thereof will be omitted. Furthermore, the selector 205 selects an abnormality detection processing mode for the calculated and grouped feature amounts, and the feature amounts are stored for each group in the storage unit 202 as detection data 207.

[0066] In step 604, the detection unit 210 determines whether an abnormal operation has occurred in the shutter device 100. Specifically, the detection unit 210 determines whether the detected data 207 stored in the storage unit 202 is abnormal data, using the abnormality detection model generated by the generation unit 209. Here, the detection unit 210 determines whether the detected data 207 stored for each group is abnormal data, using the abnormality detection model corresponding to each group. Furthermore, the detection unit 210 determines whether each of the multiple groups is abnormal data, and if at least one group is determined to contain abnormal data, it determines that the detected data 207 contains abnormal data. Furthermore, the method of abnormality detection is the same as that of the first embodiment, except that it is performed for each group, and therefore detailed description thereof will be omitted.

[0067] If it is determined in step 604 that the detected data 207 is abnormal data, the processing unit 201 proceeds to step 605, and if it is not determined that the detected data 207 is abnormal data, the processing unit 201 returns to step 603.

[0068] In step 605, the processing unit 201 acquires time-series data of physical quantities measured by the sensor 101 while the shutter device 100 is operating, and calculates feature quantities. Here, the acquisition of time-series data and calculation of feature quantities are the same as in step 601, and therefore detailed explanations will be omitted. Furthermore, the selection unit 205 selects an abnormality classification processing mode for the calculated plurality of feature quantities, and the storage unit 202 stores the calculated feature quantities as classification data 206 for classifying abnormal operations.

[0069] In step 606 , the classification unit 211 calculates first and second judgment data based on the classification data 206 .

[0070] Here, a method for calculating the first judgment data and the second judgment data will be described with reference to Fig. 8. In Fig. 8, abnormal data sets 804 to 806 are sets of data of abnormal features that belong to groups 1 to 3, respectively. Also, Mahalanobis distances 807 to 809 are the Mahalanobis distances of the abnormal data sets to the normal data sets in groups 1 to 3, respectively.

[0071] Here, when periodic abnormal operation occurs, the grouped time-series data have common characteristics, so the variance of the abnormal data set of the grouped feature quantities becomes smaller than a predetermined threshold. Periodic abnormal operation may occur, for example, when one blade deforms and comes into contact with a component of the shutter device 100. Furthermore, when periodic abnormal operation occurs, the abnormal operation occurs in a specific order of opening and closing drive within one cycle, so the Mahalanobis distance in a specific group becomes larger than the Mahalanobis distance in other groups. In the example of FIG. 8 , the Mahalanobis distance 807 of group 1 is larger than the Mahalanobis distance 808 of group 2 and the Mahalanobis distance 809 of group 3. This indicates that an abnormality has occurred in the feature quantities of group 1, and that periodic abnormal operation may be occurring in the shutter device 100.

[0072] Therefore, in this embodiment, first judgment data and second judgment data are calculated based on the variance of the feature values ​​calculated for each group and the Mahalanobis distance calculated for each group. First, the maximum value of the variance of the feature values ​​of the abnormal data set for each group is calculated as the first judgment data. Furthermore, the second judgment data is calculated as the quotient obtained by dividing the maximum value of the Mahalanobis distance for each group by the sum of the Mahalanobis distances for each group.

[0073] Returning now to the description of FIG. 6, in step 607, the classification unit 211 determines whether the first judgment data is equal to or less than a predetermined threshold. If it is determined that the first judgment data is equal to or less than the predetermined threshold, the classification unit 211 proceeds to step 608. If it is determined that the judgment data is not equal to or less than the predetermined threshold, the classification unit 211 proceeds to step 610. Here, the maximum value in the variance of the feature amount for each group is defined as max(V Gr ) and the threshold is V th Then, the judgment formula for judging the first judgment data (maximum value in the variance of the feature amount of the abnormal data set for each group) is as shown in the following formula (7).

[0074]

number

[0075] In step 607, the classification unit 211 determines whether the second judgment data is equal to or greater than a predetermined threshold. If it is determined that the second judgment data is equal to or greater than the predetermined threshold, the classification unit 211 advances the process to step 609. If it is determined that the second judgment data is not equal to or greater than the predetermined threshold, the classification unit 211 advances the process to step 610. Here, the Mahalanobis distance of group 1 is MD1, the Mahalanobis distance of group 2 is MD2, the Mahalanobis distance of group 3 is MD3, and the maximum value of the Mahalanobis distances for each group is MD4. max Let the threshold be ERR thThen, the determination formula for determining the second determination data (the quotient obtained by dividing the maximum value of the Mahalanobis distance for each group by the sum of the Mahalanobis distance for each group) is as shown in the following formula (8).

[0076]

number

[0077] In step 609, the classification unit 211 determines that a periodic abnormal operation has occurred in the shutter device 100. Furthermore, the classification unit 211 may cause the display unit 213, via the control unit 212, to display an error message indicating that a periodic abnormal operation has occurred.

[0078] In step 610, the classification unit 211 determines that a non-periodic abnormal operation has occurred in the shutter device 100. Furthermore, the classification unit 211 may cause the display unit 213, via the control unit 212, to display an error message indicating that a non-periodic abnormal operation has occurred.

[0079] As described above, according to the shutter device of this embodiment, feature amounts are calculated from the measured time-series data, and an abnormality detection model is used to determine whether an abnormal operation is occurring in the shutter device, so that abnormal operations can be detected more quickly. Furthermore, judgment data is calculated based on the feature amounts calculated for each group, and the judgment data is compared with a threshold value, so that abnormal operations occurring in the shutter device can be classified.

[0080] <Embodiments of exposure apparatus> 9 illustrates an exposure apparatus 900 incorporating a shutter apparatus 100 exemplified as the first and second embodiments. The exposure apparatus 900 exposes a substrate S coated with a photosensitive material PR by irradiating the substrate S with exposure light L1 via an original R and a projection optical system PO. The exposure apparatus 900 includes an illumination optical system 910 that illuminates the original R, and the shutter apparatus 100 is incorporated into the illumination optical system 910.

[0081] The exposure apparatus 900 can be used to manufacture articles such as semiconductor devices. An article manufacturing method according to one embodiment can include a coating process of coating a substrate S with a photosensitive material, an exposure process of exposing the substrate S that has undergone the coating process using the exposure apparatus 900, a development process of developing the photosensitive material on the substrate S that has undergone the exposure process, and a processing process of processing the substrate S that has undergone the development process. The processing process can be, for example, an ion implantation process, an etching process, or the like. The article manufacturing method can manufacture an article from the substrate S that has undergone these processes.

[0082] <Embodiment of molding device> 10 illustrates a molding apparatus 1000 incorporating a shutter device 100 exemplified as the first and second embodiments. The molding apparatus 1000 molds the composition IM on the substrate S by bringing a mold M into contact with the composition IM on the substrate S and curing the composition IM by irradiating it with light L2. The molding apparatus 1000 includes an optical system 1010 that irradiates the composition IM with light L2, and the shutter device 100 is incorporated into the optical system 1010. The molding apparatus 1000 can be configured, for example, as an imprinting apparatus that forms a pattern made of the composition IM on the substrate S, or a planarizing apparatus that forms a planarizing film made of the composition IM on the substrate S.

[0083] The molding apparatus 1000 can be used to manufacture articles such as semiconductor devices. An article manufacturing method in one embodiment includes a molding step of molding a composition IM on a substrate S using the molding apparatus 1000, and a processing step of treating the substrate S that has undergone the molding step, and an article can be manufactured from the substrate S that has undergone such steps.

[0084] <Other embodiments> The present invention can also be realized by supplying a program that realizes one or more functions of the above-described embodiments to a system or device via a network or a storage medium, and having one or more processors in the computer of the system or device read and execute the program. It can also be realized by a circuit (e.g., ASIC) that realizes one or more functions.

[0085] Although the preferred embodiments of the present invention have been described above, it goes without saying that the present invention is not limited to these embodiments, and various modifications and changes are possible within the scope of the gist of the present invention.

[0086] Furthermore, the first embodiment 1 and the second embodiment can be implemented not only independently but also in combination with each other.

Claims

1. An abnormality detection device that detects abnormal operation of a shutter device that blocks light, a detection unit that detects the abnormal operation using an abnormality detection model that receives information about measurement data of the shutter device and outputs judgment data for detecting the abnormal operation; the information about the measurement data includes a feature calculated from the measurement data, The abnormality detection device is characterized in that the feature amount includes information about a center of gravity calculated from a waveform of time-series data acquired by measuring a physical quantity related to the shutter device.

2. The abnormality detection device according to claim 1 , wherein the physical quantity includes at least one of a position, a velocity, and an acceleration of the shutter device.

3. 2. The abnormality detection device according to claim 1, wherein the physical quantity includes a sound generated from the shutter device.

4. 2. The abnormality detection device according to claim 1, wherein the physical quantity includes a torque generated by a drive mechanism of the shutter device.

5. 2. The abnormality detection device according to claim 1, wherein the physical quantity includes a current consumption consumed by a drive mechanism of the shutter device.

6. 6. The abnormality detection device according to claim 1, wherein the abnormality detection model is generated based on measurement data measured while the shutter device is operating normally.

7. 7. The abnormality detection device according to claim 1, wherein the judgment data includes a Mahalanobis distance calculated based on a feature calculated from the measurement data and a feature calculated from measurement data measured when the shutter device is operating normally.

8. The abnormality detection device according to any one of claims 1 to 7, characterized in that the detection unit determines whether the detected abnormal operation is a periodic abnormal operation based on time series data obtained by measuring physical quantities related to the shutter device, the time series data being divided according to the timing at which the shutter device drives the shutter.

9. a control unit for controlling the operation of the shutter device; The abnormality detection device according to claim 8 , wherein the control unit stops operation of the shutter device when the abnormal operation detected by the detection unit is determined to be a periodic abnormal operation.

10. The abnormality detection device described in claim 9, characterized in that when the abnormal operation detected by the detection unit is classified as a non-periodic abnormal operation, the control unit reduces the speed at which the shutter of the shutter device is driven or the frequency at which the shutter is driven and continues operating the shutter device.

11. An abnormality detection method for detecting abnormal operation of a shutter device that blocks light, comprising: a detection step of detecting the abnormal operation using an abnormality detection model that receives information about measurement data of the shutter device and outputs judgment data for detecting the abnormal operation, the information about the measurement data includes a feature calculated from the measurement data, The abnormality detection method, wherein the feature amount includes information about a center of gravity calculated from a waveform of time-series data acquired by measuring a physical quantity related to the shutter device.

12. A program for causing a computer to execute an abnormality detection method for detecting abnormal operation of a shutter device that blocks light, comprising: The abnormality detection method includes: a detection step of detecting the abnormal operation using an abnormality detection model that receives information about measurement data of the shutter device and outputs judgment data for detecting the abnormal operation, the information about the measurement data includes a feature calculated from the measurement data, The program, wherein the feature amount includes information about the center of gravity calculated from a waveform of time-series data acquired by measuring a physical amount related to the shutter device.

13. A shutter device that blocks light, a detection unit that detects the abnormal operation using an abnormality detection model that receives information related to measurement data of the shutter device and outputs judgment data for detecting the abnormal operation of the shutter device; the information about the measurement data includes a feature calculated from the measurement data, The shutter device, wherein the feature amount includes information about a center of gravity calculated from a waveform of time-series data acquired by measuring a physical quantity related to the shutter device.

14. An exposure apparatus that exposes a substrate by irradiating it with light, a shutter device that blocks the light; an abnormality detection device that detects abnormal operation of the shutter device; The abnormality detection device a detection unit that detects the abnormal operation using an abnormality detection model that receives information about measurement data of the shutter device and outputs judgment data for detecting the abnormal operation; the information about the measurement data includes a feature calculated from the measurement data, An exposure apparatus characterized in that the feature amount includes information about the center of gravity calculated from a waveform of time-series data acquired by measuring a physical amount related to the shutter device.

15. an exposure step of exposing a substrate using the exposure apparatus according to claim 14; a developing step of developing the exposed substrate; a processing step of processing the developed substrate, A method for manufacturing an article, comprising manufacturing an article from the processed substrate.

16. A molding apparatus for molding a composition by irradiating light onto a composition on a substrate in contact with a mold, the apparatus comprising: a shutter device that blocks the light; an abnormality detection device that detects abnormal operation of the shutter device; The abnormality detection device a detection unit that detects the abnormal operation using an abnormality detection model that receives information about measurement data of the shutter device and outputs judgment data for detecting the abnormal operation; the information about the measurement data includes a feature calculated from the measurement data, The molding device, wherein the feature amount includes information about the center of gravity calculated from a waveform of time-series data acquired by measuring a physical amount related to the shutter device.

17. a molding step of molding a composition on a substrate using the molding apparatus according to claim 16; a processing step of treating the substrate on which the composition has been molded, A method for manufacturing an article, comprising manufacturing an article from the processed substrate.

Images