Stage posture estimation device, conveyance device, and stage posture estimation method

Abstract

The stage posture estimation device (18) has an input data acquisition section (91) and a posture estimation section (92) that estimates a deflection angle of the stage based on input data. Thus, the deflection angle of the stage can be estimated without always providing a large-scale measurement device. In addition, the posture estimation section 92 determines one estimation result (θr) based on a plurality of temporary estimation values (θ1, θ2, θ3,...) output by a plurality of learned models (M1, M2, M3,...). Thus, high estimation accuracy can be achieved in a variety of motion conditions of the conveyance device.

Description

TECHNICAL FIELD

[0001] The present application relates to a technique of estimating a deflection angle of a stage in a conveyance device that conveys a stage of a flat plate shape by a conveyance mechanism. BACKGROUND

[0002] Conventionally, a device that performs various processes on a substrate held by a stage while conveying the stage of a flat plate shape is known. For example, Patent Literature 1 describes a device that moves a stage (10) on which a substrate (W) is placed by a stage moving mechanism (20) while drawing an exposure pattern on the upper surface of the substrate (W).

[0003] PRIOR ART DOCUMENTS

[0004] PATENT LITERATURE

[0005] Patent Literature 1: Japanese Patent Application Publication No. 2016-72434 SUMMARY

[0006] PROBLEMS TO BE SOLVED BY THE INVENTION

[0007] A conveyance device that mounts a stage on such a device sometimes has a pair of straight line mechanisms. Specifically, a mechanism that conveys a stage in a prescribed direction by a pair of linear motors arranged parallel to each other is known.

[0008] In this conveyance device, in order to move the stage in a certain posture, it is necessary to make the pair of straight line mechanisms act equally. However, sometimes a slight driving error of the pair of straight line mechanisms, a variation in air pressure within a gap of a guide of the linear motor, a machining error, or the like causes a slight variation in the rotation angle of the stage around the vertical axis (so-called "deflection angle"). When such a variation in the deflection angle occurs, it is difficult to perform a precise process on the substrate held by the stage.

[0009] In the conventional conveyance device, a large-scale measurement device is mounted to grasp the variation in the deflection angle described above. Moreover, based on the measurement result of the measurement device, the operation of the conveyance device is corrected. However, if a large-scale measurement device is mounted, it is difficult to downsize the conveyance device. In addition, by mounting the measurement device, the manufacturing cost of the conveyance device also increases.

[0010] In order to grasp the deflection angle of the stage without always providing such a large-scale measurement device, for example, it is considered to utilize machine learning. Specifically, it is considered to prepare a learned model that has learned the relationship between a measurement value such as a torque value output by the conveyance device and the deflection angle of the stage, input the measurement value to the learned model, and output the deflection angle of the stage.

[0011] However, there are various machine learning algorithms used to generate fully learned models. Furthermore, each algorithm has its own strengths and weaknesses, and the estimation accuracy varies depending on the operation of the conveyor. Therefore, relying solely on a single fully learned model generated by a single machine learning algorithm may not always result in highly accurate estimations based on the conveyor's operational status.

[0012] The present invention was made in view of this situation, and its object is to provide a technique that can estimate the deflection angle of the stage without always setting up such a large-scale measuring device, and can achieve high estimation accuracy under various operating conditions of the conveying device.

[0013] means for solving problems

[0014] To address the aforementioned issues, the first invention of this application is a stage posture estimation device for estimating the deflection angle of a stage in a conveying device that conveys a flat stage via a conveying mechanism. The device comprises: an input data acquisition unit that acquires a measurement value output by the conveying mechanism or a value calculated based on the measurement value as input data; and a posture estimation unit that estimates the deflection angle of the stage based on the input data and outputs an estimation result. The posture estimation unit includes: multiple learned models that output temporary estimated values ​​of the deflection angle of the stage based on the input data; and an estimation value determination unit that determines the estimation result based on the multiple temporary estimated values ​​output by the multiple learned models.

[0015] The second invention of this application is a stage posture estimation device as described in the first invention, wherein the plurality of learned models are generated by different algorithms.

[0016] The third invention of this application is a stage posture estimation device as described in the first or second invention, wherein the estimation value determination unit takes the average of the plurality of temporary estimation values ​​as the estimation result.

[0017] The fourth invention of this application is a stage posture estimation device as described in the first or second invention, wherein a weighting ratio is set in the plurality of learned models, and the estimation value determination unit uses the weighted average of the plurality of temporary estimation values ​​using the weighting ratio as the estimation result.

[0018] The fifth invention of this application is a stage posture estimation device as described in the fourth invention, wherein the estimation value determination unit changes the weighting ratio based on a state variable representing the operating state of the conveying mechanism.

[0019] The sixth invention of this application is a platform posture estimation device as described in the first or second invention, wherein the estimation value determination unit selects one of the plurality of temporary estimation values ​​based on a state variable representing the operating state of the conveying mechanism, and uses the selected temporary estimation value as the estimation result.

[0020] The seventh invention of this application is a stage posture estimation device as described in any one of the first to sixth inventions, wherein the conveying mechanism conveys the stage by a pair of straight-moving mechanisms, and the input data acquisition unit generates the input data based on the difference in torque values ​​of the pair of straight-moving mechanisms.

[0021] The eighth invention of this application is a conveying device having a stage posture estimation device as described in any one of the first to seventh inventions; the stage; and the conveying mechanism.

[0022] The ninth invention of this application is a method for estimating the deflection angle of a flat platform in a conveying device that conveys a flat platform by a conveying mechanism, comprising: step a), acquiring a measurement value output by the conveying mechanism or a value calculated based on the measurement value as input data; and step b), estimating the deflection angle of the platform based on the input data and outputting the estimation result, wherein in step b), the input data is input into multiple learned models, and the estimation result is determined based on multiple provisional estimation values ​​output by the multiple learned models.

[0023] The effects of the invention

[0024] According to the first to ninth inventions of this application, the deflection angle of the stage is estimated based on the measured value output by the conveying mechanism or the value calculated based on the measured value. Therefore, the deflection angle of the stage can be estimated without always having a large-scale measuring device. Furthermore, an estimation result is output based on multiple provisional estimation values ​​output by multiple learned models. Thus, high estimation accuracy can be achieved under various operating conditions of the conveying mechanism. Attached Figure Description

[0025] Figure 1 It is a three-dimensional drawing of a device with a conveying mechanism.

[0026] Figure 2 It is a schematic top view of the device with a conveying mechanism.

[0027] Figure 3 It is a block diagram showing the electrical connections between the control unit and the various parts within the drawing device.

[0028] Figure 4 This is a partial cross-sectional view when a portion of the conveying device is cut off by a plane perpendicular to the main scanning direction.

[0029] Figure 5 This is a block diagram showing the structure of the stage posture estimation device.

[0030] Figure 6 It is a flowchart representing the pre-learning and processing flow.

[0031] Figure 7 This is a flowchart illustrating the estimation process of the stage posture estimation device.

[0032] Figure 8 This is a flowchart of a second example of a method for determining the estimation result of the estimation value determination unit.

[0033] Figure 9 This is a flowchart of the third example of a method for determining the estimation result of the estimation value determination unit.

[0034] Figure 10 This is a flowchart illustrating the fourth example of a method for determining the estimation result of the estimation value determination unit. Detailed Implementation

[0035] The embodiments of the present invention will now be described with reference to the accompanying drawings.

[0036] It should be noted that, in the following, the direction in the horizontal direction in which the stage is moved by a pair of straight-line mechanisms is defined as the "main scanning direction", and the direction orthogonal to the main scanning direction is defined as the "secondary scanning direction".

[0037] <1. Describe the structure of the device>

[0038] Figure 1 This is a perspective view of a drawing device 1 having a conveying device 10 according to an embodiment of the present invention. Figure 2 This is a schematic top view of the drawing apparatus 1. The drawing apparatus 1 is a device that illuminates the upper surface of a substrate W, such as a semiconductor substrate or a glass substrate coated with a photosensitive material, with spatially modulated light, thereby drawing an exposure pattern on the upper surface of the substrate W. Figure 1 and Figure 2 As shown, the drawing device 1 includes a conveying device 10, a frame 20, a drawing processing unit 30, and a control unit 40.

[0039] The transfer device 10 is a means of horizontally transferring a flat stage 12 in a generally defined posture on the upper surface of the base 11. The transfer device 10 has a transfer mechanism including a main scanning mechanism 13 and a sub-scanning mechanism 14. The main scanning mechanism 13 is used to transfer the stage 12 along the main scanning direction. The sub-scanning mechanism 14 is used to transfer the stage 12 along the sub-scanning direction. The substrate W is held horizontally on the upper surface of the stage 12 and moves together with the stage 12 along the main scanning direction and the sub-scanning direction.

[0040] The more detailed structure of the conveying device 10 is described below.

[0041] The frame 20 is a structure for holding the drawing processing unit 30 above the base 11. The frame 20 has a pair of support sections 21 and a bridging section 22. The pair of support sections 21 are spaced apart in the sub-scanning direction. Each support section 21 extends upward from the upper surface of the base 11. The bridging section 22 extends in the sub-scanning direction between the upper ends of the two support sections 21. The stage 12 holding the substrate W can pass between the pair of support sections 21 and below the bridging section 22.

[0042] The drawing processing unit 30 includes two optical heads 31, an illumination optical system 32, a laser oscillator 33, and a laser drive unit 34. The two optical heads 31 are fixed to the bridging unit 22 at a distance from each other in the sub-scanning direction. The illumination optical system 32, the laser oscillator 33, and the laser drive unit 34 are housed, for example, within the internal space of the bridging unit 22. The laser drive unit 34 is electrically connected to the laser oscillator 33. When the laser drive unit 34 is activated, pulsed light is emitted from the laser oscillator 33. Furthermore, the pulsed light emitted from the laser oscillator 33 is guided into the optical heads 31 through the illumination optical system 32.

[0043] An optical system including a spatial modulator is provided inside the optical head 31. For example, a GLV (Grating Light Valve) (registered trademark) is used as a diffraction grating type spatial light modulator. Pulsed light introduced into the optical head 31 is modulated into a predetermined pattern by the spatial modulator and irradiates the upper surface of the substrate W. As a result, photosensitive materials such as resist coated on the upper surface of the substrate W are exposed.

[0044] When the drawing apparatus 1 is operating, the exposure of the optical head 31 and the transport of the substrate W by the transport device 10 are repeatedly performed. Specifically, while the stage 12 is transported along the sub-scanning direction by the sub-scanning mechanism 14, pulsed light from the optical head 31 is applied to expose a strip-shaped area (swath) extending along the sub-scanning direction. Then, the stage 12 is transported along the main scanning direction by the main scanning mechanism 13 in the amount of one swath. By repeatedly performing this exposure in the sub-scanning direction and transport of the stage 12 in the main scanning direction, the drawing apparatus 1 draws a pattern on the entire upper surface of the substrate W.

[0045] The control unit 40 is a mechanism for controlling the operation of various parts of the drawing device 1. Figure 3 This is a block diagram showing the electrical connections between the control unit 40 and the various parts within the drawing device 1. For example... Figure 3As schematically shown, the control unit 40 is composed of a computer having a processor 41 such as a CPU, a memory 42 such as RAM, and a storage unit 43 such as a hard disk drive. The storage unit 43 stores a computer program P for controlling the operation of the drawing device 1.

[0046] In addition, such as Figure 3 As shown, the control unit 40 is electrically connected to the drawing processing unit 30 (including the aforementioned optical head 31 and laser drive unit 34), the main scanning mechanism 13 (including the linear motor 61 and air guide 62 described later), the sub-scanning mechanism 14 (including the linear motor 71 described later), the rotation mechanism 15 described later, and various sensors 50. The control unit 40 reads the computer program P and data D stored in the storage unit 43 into the memory 42. The processor 41 performs calculations based on the computer program P and data D, thereby controlling the operation of the aforementioned parts within the drawing device 1. Thus, drawing processing is performed in the drawing device 1.

[0047] <2. Structure of the conveying device>

[0048] The detailed structure of the conveying device 10 will be described below. Figure 4 This is a partial cross-sectional view when a portion of the conveying device is cut off by a plane perpendicular to the main scanning direction. For example... Figures 1 to 4 As shown, the conveying device 10 includes a base 11, a stage 12, a main scanning mechanism 13, a secondary scanning mechanism 14, a rotating mechanism 15, a lower support plate 16, a middle support plate 17, and a stage posture estimation device 18.

[0049] The base 11 is a support platform that supports the various parts of the conveying device 10. The base 11 has a flat, plate-like shape that extends along the main scanning direction and the sub-scanning direction. Four feet 111 and two dampers 112 are provided on the lower surface of the base 11. The lengths of the feet 111 and the dampers 112 can be adjusted individually. Therefore, by adjusting the lengths of the feet 111 and the dampers 112, the posture of the base 11 can be adjusted to be horizontal.

[0050] The lower support plate 16, the middle support plate 17, and the stage 12 each have a flat, plate-like shape. The lower support plate 16 is supported on the base 11 and is movable along the main scanning direction via the main scanning mechanism 13. The middle support plate 17 is supported on the lower support plate 16 and is movable along the sub-scanning direction via the sub-scanning mechanism 14. The stage 12 is supported on the middle support plate 17 and is rotatable about a vertical axis via a rotation mechanism 15. The stage 12 has an upper surface capable of holding the substrate W. Furthermore, the upper surface of the stage 12 is provided with chuck pins for holding the substrate W or multiple suction holes for adsorbing the substrate W.

[0051] The main scanning mechanism 13 is a mechanism that moves the lower support plate 16 relative to the base 11 along the main scanning direction. The main scanning mechanism 13 has a pair of linear mechanisms 60. The pair of linear mechanisms 60 are disposed at both ends of the upper surface of the base 11 in the secondary scanning direction. Figure 2 and Figure 4 As shown, a pair of straight-line mechanisms 60 each have a linear motor 61 and an air guide 62.

[0052] The linear motor 61 has a stator 611 and a mover 612. The stator 611 is laid on the upper surface of the base 11 along the main scanning direction. That is, a pair of stators 611 are arranged parallel to each other. The mover 612 is fixed to the lower support plate 16 by an air bearing 622 described later.

[0053] Additionally, the main scanning mechanism 13 has a control board 63 for controlling the operation of the linear motor 61. For example, Servopack (registered trademark) is used as the control board 63. The control board 63 is electrically connected to the control unit 40. When driving the linear motor 61, the control board 63 calculates the torque to be generated in the linear motor 61 according to the instructions from the control unit 40. Furthermore, a drive signal corresponding to the calculated torque is supplied to the stator 611 of each linear motor 61. In this way, utilizing the magnetic attraction and reaction forces generated between the stator 611 and the mover 612, the mover 612 moves along the stator 611 in the main scanning direction.

[0054] The air guide 62 includes a guide rail 621 and an air bearing 622. The guide rail 621 is laid on the upper surface of the base 11 along the main scanning direction. That is, the stator 611 of the linear motor 61 and the guide rail 621 of the air guide 62 are arranged parallel to each other. The air bearing 622 is fixed to the lower support plate 16 and the mover 612. In addition, the air bearing 622 is disposed above the guide rail 621.

[0055] like Figure 4 As shown, a gas outlet 623 is provided on the lower surface of the air bearing 622. During operation of the conveying device 10, gas is continuously supplied to the air bearing 622 from the equipment in the factory, and pressurized gas is blown out from the gas outlet 623 onto the upper surface of the guide rail 621. Thus, the air bearing 622 is non-contactly supported on the guide rail 621. Therefore, when the linear motor 61 is driven, the lower support plate 16 moves smoothly along the main scanning direction with low friction while being supported by the air guide 62.

[0056] The sub-scanning mechanism 14 is a mechanism that moves the middle support plate 17 relative to the lower support plate 16 along the sub-scanning direction. The sub-scanning mechanism 14 has a linear motor 71 and a pair of guide mechanisms 72.

[0057] A linear motor 71 is positioned approximately at the center of the main scanning direction on the upper surface of the lower support plate 16. The linear motor 71 has a stator 711 and a mover 712. The stator 711 is laid on the upper surface of the lower support plate 16 along the secondary scanning direction. The mover 712 is fixed to the middle support plate 17. When the linear motor 71 is driven, the mover 712 moves along the stator 711 in the secondary scanning direction using the attractive and reactive forces generated between the stator 711 and the mover 712.

[0058] A pair of guide mechanisms 72 are disposed at both ends of the upper surface of the lower support plate 16 in the main scanning direction. Each guide mechanism 72 has a guide rail 721 and a ball bearing 722. The guide rail 721 is laid along the sub-scanning direction on the upper surface of the lower support plate 16. The ball bearing 722 is fixed to the lower surface of the middle support plate 17. Furthermore, the ball bearing 722 is movable along the guide rail 721 in the sub-scanning direction. Therefore, when the linear motor 71 is driven, the middle support plate 17 moves relative to the lower support plate 16 in the sub-scanning direction.

[0059] The rotating mechanism 15 is a mechanism for adjusting the angle of the stage 12 relative to the middle support plate 17 about the vertical axis. For example, a motor is used as the rotating mechanism 15. When the motor is actuated, the stage 12 rotates about the vertical axis relative to the middle support plate 17. This allows adjustment of the angle (deflection angle) θ of the stage 12 about the vertical axis.

[0060] In this way, the stage 12 can move relative to the base 11 along the main scanning direction and the sub-scanning direction through the main scanning mechanism 13, the sub-scanning mechanism 14 and the rotating mechanism 15, and can adjust the deflection angle θ.

[0061] like Figure 1 and Figure 2 As shown, an attitude measuring device 80 can be installed in the conveying device 10. The attitude measuring device 80 is a device for measuring the deflection angle θ of the stage 12. The attitude measuring device 80 includes a reflector 81 fixed to the stage 12 and a laser interferometer 82. The reflector 81 is fixed to the end edge of the stage 12 in the main scanning direction. The laser interferometer 82 is fixed to the upper surface of the base 11. The laser interferometer 82 irradiates two lasers onto the reflector 81. Furthermore, the optical path difference between the two lasers is detected based on the interference of the two lasers reflected from the reflector 81. Then, the deflection angle θ of the stage 12 is measured based on this optical path difference.

[0062] The posture measuring device 80 is set up after the prior learning process described later. After the prior learning is completed, the posture measuring device 80 can be removed and the conveying device 10 can be used.

[0063] <3. Regarding the stage attitude estimation device>

[0064] <3-1. Structure of the Stage Attitude Estimation Device>

[0065] Next, the posture estimation device 18 of the platform mounted on the conveying device 10 will be described. Figure 5 This is a block diagram showing the structure of the stage posture estimation device 18. The stage posture estimation device 18 is a device that estimates the deflection angle θ of the stage 12 based on the measurement values ​​output from the main scanning mechanism 13. (Example...) Figure 5 As shown, the stage posture estimation device 18 includes an input data acquisition unit 91 and a posture estimation unit 92. The input data acquisition unit 91 includes a measurement value input unit 911 and an input data generation unit 912. The posture estimation unit 92 includes multiple learned models M1, M2, M3... and an estimation value determination unit 921.

[0066] The stage posture estimation device 18 is composed of a computer having a processor such as a CPU, a memory such as RAM, and a storage unit such as a hard disk drive. The functions of the measurement value input unit 911, the input data generation unit 912, and the estimation value determination unit 921 are implemented by the processor according to the computer program stored in the storage unit.

[0067] The learned models M1, M2, M3... are inference procedures whose parameters have been adjusted through prior learning using machine learning algorithms. The machine learning algorithms used to obtain the learned models M1, M2, M3... include, for example, neural networks such as single-layer neural networks and deep learning, decision tree algorithms such as random forests and gradient-driven algorithms, and so-called supervised machine learning algorithms such as support vector machines. Multiple learned models M1, M2, M3... are generated by different machine learning algorithms.

[0068] It should be noted that the stage posture estimation device 18 can be composed of the same computer as the control unit 40, or it can be composed of a different computer from the control unit 40.

[0069] <3-2. Prior learning of the stage posture estimation device>

[0070] Next, the pre-learning process performed in the stage posture estimation device 18 will be explained. Figure 6 It is a flowchart representing the pre-learning and processing flow.

[0071] During the pre-learning process, the posture measuring device 80 is installed in the conveying device 10 (step S11). Furthermore, the following steps S12 to S15 are repeatedly performed while the main scanning mechanism 13 is operated.

[0072] First, the measurement value output by the control board 63 is input into the measurement value input unit 911 (step S12). The measurement value is, for example, the torque value of a pair of linear motors 61 of the main scanning mechanism 13. However, the measurement value input to the measurement value input unit 911 may also be a value obtained by measuring other items such as the air pressure of the air bearing 622, the temperature of the guide rail 621, the drive sound of the conveying device 10, the vibration of the stage 12, and the position of the stage 12 through various sensors 50.

[0073] Then, the input data generation unit 912 generates input data d based on the measured values ​​input to the measurement value input unit 911 (step S13). For example, if the measured value is the torque value of a pair of linear motors 61, the input data generation unit 912 calculates the difference between these torque values. Furthermore, unwanted frequencies are removed from the time series data of the calculated difference to generate input data d. If the measured value input to the measurement value input unit 911 is not a torque value, the input data generation unit 912 also performs prescribed calculations and filtering to generate input data d suitable for machine learning.

[0074] Furthermore, the input data generation unit 912 can also use the measurement value itself input to the measurement value input unit 911 as input data d. That is, the input data generation unit 912 can simply set the measurement value output by the conveying mechanism or the value calculated by performing a prescribed operation and filtering on the measurement value as input data d.

[0075] Next, the posture estimation unit 92 takes the input data d generated by the input data generation unit 912 as input and uses the measurement result θm of the posture measuring device 80 as the training dataset to perform machine learning (step S14). That is, the posture estimation unit 92 learns the relationship between the input data d and the measurement result θm of the posture measuring device 80 through the above-described machine learning algorithm. In this embodiment, the posture estimation unit 92 performs the machine learning in step S14 in parallel using multiple different machine learning algorithms. Therefore, multiple different learned models M1, M2, M3, etc. are generated through the machine learning in step S14.

[0076] The posture estimation unit 92 compares the output values ​​of the learned models M1, M2, M3... generated through machine learning in step S14 with the measurement results θm, which serve as the training dataset. Furthermore, if the difference between the output values ​​of the learned models M1, M2, M3... and the measurement results θm is not below a preset threshold, the posture estimation unit 92 determines that the estimation accuracy of the learned models M1, M2, M3... has not reached the desired level (step S15: No). In this case, the stage posture estimation device 18 repeatedly performs the above steps S12 to S14. Thus, through repeated machine learning, the estimation accuracy of the learned models M1, M2, M3... gradually improves.

[0077] Finally, when the difference between the output values ​​of the learned models M1, M2, M3... and the measurement result θm is below a preset threshold, the posture estimation unit 92 determines that the estimation accuracy of each learned model M1, M2, M3... has reached the desired level. In this case, the posture estimation unit 92 ends the machine learning (step S15: Yes). Furthermore, the posture measurement device 80 is removed from the conveying device 10 (step S16). It should be noted that when the number of repetitions of steps S12 to S14 reaches a preset upper limit, the stage posture estimation device 18 can end the machine learning in step S15.

[0078] <3-3. Estimation Process of Stage Attitude Estimation Device>

[0079] Next, the estimation process based on the deflection angle θ of the stage posture estimation device 18 will be explained. After the above-mentioned pre-learning process is completed, this estimation process is performed when the conveying device 10 is operated. Figure 7 This is a flowchart representing the presumed processing flow.

[0080] During the estimation process, firstly, the measurement value output by the control board 63 is input to the measurement value input unit 911 (step S21). Here, the same type of measurement value as in step S12 is input. For example, if the measurement value input in step S12 is the torque value of a pair of linear motors 61, the measurement value input in step S21 is also the torque value of a pair of linear motors 61.

[0081] Then, the input data generation unit 912 generates input data d based on the measurement value input to the measurement value input unit 911 (step S22). Here, the same processing as in step S13 is performed. That is, if the processing performed in step S13 is the calculation of the difference and the filtering process, the calculation of the difference and the filtering process are also performed in step S22 to generate the input data d.

[0082] Next, the posture estimation unit 92 inputs the generated input data d into multiple learned models M1, M2, M3... (step S23). In this way, each learned model M1, M2, M3... outputs a temporary estimated value θ1, θ2, θ3... of the deflection angle θ of the stage 12 corresponding to the input data d. Thus, for a single input data d, multiple temporary estimated values ​​θ1, θ2, θ3... are obtained (step S24).

[0083] Next, the estimation value determination unit 921 of the attitude estimation unit 92 determines an estimation result θr based on multiple temporary estimation values ​​θ1, θ2, θ3... (step S25). Specifically, for example, the estimation value determination unit 921 calculates the average value of the multiple temporary estimation values ​​θ1, θ2, θ3... and determines the calculated average value as the estimation result θr. However, the estimation value determination unit 921 may also determine the estimation result θr using other methods.

[0084] Figure 8 This is a flowchart illustrating a second example of the method for determining the estimation result of the estimation value determination unit 921. Figure 8 In the example, weighting ratios w1, w2, w3, etc., are pre-set for the learned models M1, M2, M3, etc. These weighting ratios w1, w2, w3, etc., are pre-stored in the storage unit of the computer constituting the stage posture estimation device 18. First, the estimation value determination unit 921 reads the weighting ratios w1, w2, w3, etc., from the storage unit (step S31). Then, the estimation value determination unit 921 uses the read weighting ratios w1, w2, w3, etc., to calculate a weighted average of multiple temporary estimation values ​​θ1, θ2, θ3, etc., and sets the calculated weighted average as the estimation result θr (step S32).

[0085] For example, when using three learned models M1, M2, and M3, the weighted average estimate θr in step S32 can be calculated using the following equation (1).

[0086] θr=(w1·θ1+w2·θ2+w3·θ3) / (w1+w2+w3) (1)

[0087] Among multiple learned models M1, M2, M3..., when there is a learned model with particularly high estimation accuracy or a learned model that deserves attention, it is preferable to set the weighting ratio of that learned model relatively high. In this way, by calculating the weighted average as described above, a more preferred estimation result θr can be obtained.

[0088] Figure 9 This is a flowchart illustrating a third example of the method for determining the estimation result of the estimation value determination unit 921. Figure 9 In the example, the weighting ratios w1, w2, w3... corresponding to the learned models M1, M2, M3... are not fixed values, but vary according to the operating state of the main scanning mechanism 13. The estimation value determination unit 921 first acquires the state variable representing the operating state of the main scanning mechanism 13 (step S41). The state variable can be the measurement value input to the measurement value input unit 911, the variable acquired by other sensors, or the operating mode set by the user for the control unit 40, etc.

[0089] The estimation value determination unit 921 changes the weighting ratios w1, w2, w3... based on the acquired state variables (step S42). Thus, the weighting ratios of the learned models that achieve high estimation accuracy are increased according to the operating state of the main scanning mechanism 13. For example, in an operating state represented by a certain state variable, if the estimation accuracy of the learned model M2 among the multiple learned models M1, M2, M3... is particularly high, the estimation value determination unit 921 changes the multiple weighting ratios w1, w2, w3... to make the value of the weighting ratio w2 relatively high.

[0090] Then, the estimation value determination unit 921 uses the changed weighting ratios w1, w2, w3... to calculate the weighted average of multiple temporary estimation values ​​θ1, θ2, θ3... and uses the calculated weighted average as the estimation result θr (step S43).

[0091] Thus, by changing the weighting ratios w1, w2, w3, etc., according to the action state, the weighting of the learned model can be adjusted for each action state. Therefore, by increasing the weighting ratios of the learned model to achieve high inference accuracy for each action state, a more accurate inference result θr can be obtained.

[0092] Figure 10 This is a flowchart illustrating the fourth example of a method for determining the estimation result of the estimation value determination unit 921. Figure 10 In the example, the estimation value determination unit 921 first acquires a state variable representing the operating state of the main scanning mechanism 13 (step S51). The state variable can be a measurement value input to the measurement value input unit 911, a variable acquired by other sensors, or an operating mode set by the user for the control unit 40, etc.

[0093] The estimation value determination unit 921 selects one of a plurality of temporary estimation values ​​θ1, θ2, θ3... based on the acquired state variable, and uses the selected temporary estimation value as the estimation result θr (step S52). For example, in an action state represented by a certain state variable, if the estimation accuracy of the learned model M2 among the plurality of learned models M1, M2, M3... is particularly high, the estimation value determination unit 921 uses the temporary estimation value θ2 output by the learned model M2 as the estimation result θr.

[0094] In this example, based on the action state, one of the multiple temporary estimated values ​​θ1, θ2, θ3, etc., output by multiple learned models M1, M2, M3, etc., is adopted. In this way, a learned model that can achieve high estimation accuracy is selected for each action state, thereby obtaining a high-precision estimation result θr.

[0095] Return to Figure 7When the estimated result θr of the deflection angle θ is determined, the estimation value determination unit 921 outputs the estimated result θr to the control unit 40 (step S26). Next, the control unit 40 corrects the deflection angle θ of the stage 12 based on the estimated result θr of the deflection angle θ output by the posture estimation unit 92 (step S27). Specifically, the control unit 40 actuates the rotation mechanism 15 or adjusts the torque value of one of the pair of linear motors 61. As a result, the deflection angle θ of the stage 12 is corrected to a value close to the desired value.

[0096] As described above, in this conveying device 10, the deflection angle θ of the stage 12 is estimated based on input data d, which is a measured value output by the main scanning mechanism 13 or a value calculated based on the measured value. Therefore, the deflection angle θ of the stage 12 can be estimated without always having a large-scale posture measuring device 80 installed.

[0097] Furthermore, in this conveying device 10, a single estimation result θr is output based on multiple temporary estimation values ​​θ1, θ2, θ3, etc., output by multiple learned models M1, M2, M3, etc. Therefore, high estimation accuracy can be achieved under various operating conditions of the conveying device 10. In this embodiment, the multiple learned models M1, M2, M3, etc., are each generated by different machine learning algorithms. Therefore, in operating conditions where a particular machine learning algorithm struggles to achieve high estimation accuracy, other machine learning algorithms can supplement it. Thus, the deflection angle θ of the stage 12 can be estimated with high accuracy under a wider range of operating conditions.

[0098] <4. Variations>

[0099] The above describes one embodiment of the present invention, but the present invention is not limited to the above embodiment.

[0100] In the above embodiment, the measurement value input unit 911 obtains the torque value of the linear motor 61 from the control board 63. However, the measurement value input unit 911 can also obtain the torque value of the linear motor 61 by other methods. For example, a torque sensor can be installed on each linear motor 61 of the straight-line mechanism 60 and the torque value can be obtained from the torque sensor.

[0101] Furthermore, in the above embodiment, the same input data d is input to all learned models M1, M2, M3... However, the input data generation unit 912 may also perform different processing on the measurement values ​​input to the measurement value input unit 911 for each learned model. Moreover, different input data generated by different processing may be input to multiple learned models M1, M2, M3... In this way, more appropriate input data can be input to each learned model.

[0102] Furthermore, the conveying device 10 of the above embodiment not only has a main scanning mechanism 13, but also a sub-scanning mechanism 14 and a rotating mechanism 15. However, the present invention can also be applied to a conveying device that does not have a sub-scanning mechanism 14 and a rotating mechanism 15.

[0103] Furthermore, the conveying device 10 of the above embodiment is mounted on the drawing device 1. However, the present invention can also be applied to a conveying device mounted on a device other than the drawing device 1. For example, the conveying device can also be mounted on a device for applying a processing liquid to a substrate held on a stage. In addition, the conveying device can also be mounted on a device for printing on a recording medium held on a stage.

[0104] Furthermore, in the above embodiment, the deflection angle θ of the stage 12 is measured by the laser interferometer 82 during the prior learning process. However, the deflection angle θ of the stage 12 can also be measured by other methods. For example, the deflection angle θ of the stage 12 can also be measured based on an image of the stage 12 acquired by a camera.

[0105] Furthermore, the linear mechanism 60 in the above embodiment includes a linear motor 61. However, a mechanism that converts the rotational motion output by a rotary motor into linear motion via a ball screw can be used instead of the linear motor 61.

[0106] Furthermore, without creating contradictions, the elements appearing in the above-described implementation methods and variations can be appropriately combined.

[0107] Explanation of reference numerals in the attached figures

[0108] 1: Drawing device

[0109] 10: Transport device

[0110] 11: Abutment

[0111] 12: Platform

[0112] 13: Main scanning mechanism

[0113] 14: Sub-scanning mechanism

[0114] 15: Rotating mechanism

[0115] 16: Lower support plate

[0116] 17: Middle layer support plate

[0117] 18: Stage attitude estimation device

[0118] 20: Framework

[0119] 30: Drawing Processing Department

[0120] 40: Control Department

[0121] 50: Sensor

[0122] 60: Straight-through agency

[0123] 61: Linear Motor

[0124] 62: Air guide

[0125] 63: Control board

[0126] 80: Posture measuring device

[0127] 91: Input Data Acquisition Department

[0128] 92: Posture estimation section

[0129] 911: Measurement Input Section

[0130] 912: Input Data Generation Department

[0131] 921: Estimated Value Determination Section

[0132] M1, M2, M3: Models that have completed learning

[0133] W: substrate

[0134] D: Input data

[0135] θ: Deflection angle

[0136] θ1, θ2, θ3: Provisional estimated values

[0137] θr: Estimated result

Claims

1. A stage posture estimation device, used in a conveying device that conveys a flat stage via a conveying mechanism to estimate the deflection angle of the stage, wherein, have: The input data acquisition unit acquires the measurement value output by the conveying mechanism or the value calculated based on the measurement value as input data; as well as The attitude estimation unit estimates the deflection angle of the stage based on the input data and outputs the estimation result. The posture estimation unit has: Multiple learned models output a provisional estimate of the deflection angle of the stage based on the input data; as well as The estimation value determination unit determines the estimation result based on the multiple provisional estimation values ​​output by the multiple learned models. A weighted ratio is set among the multiple learned models. The estimation value determination unit changes the weighting ratio based on the state variable representing the operating state of the conveying mechanism. The estimation value determination unit uses the weighted average of the multiple provisional estimation values ​​that use the weighting ratio as the estimation result.

2. A stage posture estimation device, used in a conveying device that conveys a flat stage via a conveying mechanism, to estimate the deflection angle of the stage, wherein, have: The input data acquisition unit acquires the measurement value output by the conveying mechanism or the value calculated based on the measurement value as input data; as well as The attitude estimation unit estimates the deflection angle of the stage based on the input data and outputs the estimation result. The posture estimation unit has: Multiple learned models output a provisional estimate of the deflection angle of the stage based on the input data; as well as The estimation value determination unit determines the estimation result based on the multiple provisional estimation values ​​output by the multiple learned models. The estimation value determination unit selects one of the multiple temporary estimation values ​​based on the state variable representing the operation state of the conveying mechanism, and uses the selected temporary estimation value as the estimation result.

3. The stage posture estimation device as described in claim 1 or 2, wherein, The multiple learned models are generated by different algorithms.

4. The stage posture estimation device as described in claim 1 or 2, wherein, The conveying mechanism moves the platform via a pair of straight-line mechanisms. The input data acquisition unit generates the input data based on the difference in torque values ​​of the pair of straight-line mechanisms.

5. A conveying device, wherein, have: The stage posture estimation device according to claim 1 or 2; The stage; and The conveying mechanism.

6. A method for estimating the posture of a platform, wherein the deflection angle of the platform is estimated in a conveying device that conveys a flat platform via a conveying mechanism, wherein, include: Step a) Obtain the measured value output by the conveying mechanism or the value calculated based on the measured value as input data; as well as Step b) Based on the input data, multiple learned models are used to estimate the deflection angle of the stage, and the estimation results are output. A weighted ratio is set among the multiple learned models. The weighting ratio changes based on a state variable representing the operational state of the conveying mechanism. In step b), the input data is input into the plurality of learned models to obtain a plurality of temporary estimated values ​​output by the plurality of learned models, and the weighted average of the plurality of temporary estimated values ​​using the weighting ratio is taken as the estimated result.

7. A method for estimating the posture of a platform, wherein the deflection angle of the platform is estimated in a conveying device that conveys a flat platform via a conveying mechanism, wherein, include: Step a) Obtain the measured value output by the conveying mechanism or the value calculated based on the measured value as input data; as well as Step b) Based on the input data, multiple learned models are used to estimate the deflection angle of the stage, and the estimation results are output. In step b), the input data is input into the plurality of learned models, a plurality of temporary estimated values ​​output by the plurality of learned models are obtained, one of the plurality of temporary estimated values ​​is selected based on the state variable representing the action state of the conveying mechanism, and the selected temporary estimated value is used as the estimated result.

Images