Information processing method, information processing device, and computer program

Abstract

Provided are an information processing method, an information processing device, and a computer program for easily realizing a base model for estimating information relating to a substrate processing device.　In the information processing device: data relating to the substrate processing device are input into a first model, and a first feature quantity output by the first model is acquired; a query relating to the substrate processing device is input into a second model, and a second feature quantity output by the second model is acquired; the first feature quantity and the second feature quantity are input into an integrated model, and an answer to the query output by the integrated model is acquired; and the acquired answer is output.

Description

Information processing method, information processing device, and computer program

[0001] The present disclosure relates to an information processing method, an information processing device, and a computer program.

[0002] There is a need to infer some information that is not available about a substrate processing apparatus from some data obtained about the substrate processing apparatus that performs processes such as etching or film formation on substrates such as semiconductor wafers or glass substrates. For example, it may be desirable to infer the outcome of substrate processing from log data showing the history of the state of the substrate processing apparatus and data showing the state of the substrate before processing. In order to build a base model for making inferences, it is conceivable to use a large language model (LLM). Non-Patent Document 1 discloses an application of an LLM.

[0003] A Survey of Large Language Models, 2023

[0004] Conventionally, a base model has been required for each substrate processing apparatus, or the base model has had to be tuned for each substrate processing apparatus, which can make development or operation of the base model difficult.

[0005] The present disclosure provides an information processing method, an information processing device, and a computer program for easily realizing a base model for estimating information related to a substrate processing device.

[0006] An information processing method according to one aspect of the present disclosure inputs data related to a substrate processing apparatus into a first model, acquires a first feature output by the first model, inputs a query related to the substrate processing apparatus into a second model, acquires a second feature output by the second model, inputs the first feature and the second feature into an integrated model, acquires an answer to the query output by the integrated model, and outputs the acquired answer.

[0007] According to the present disclosure, it is possible to provide an information processing method, an information processing apparatus, and a computer program for easily realizing a base model for estimating information related to a substrate processing apparatus.

[0008] 1 is a conceptual diagram showing an example of the configuration of a base model. FIG. 2 is a diagram showing an example of the contents of log data. FIG. 3 is a diagram showing an example of performance data. FIG. 4 is a block diagram showing an example of the internal configuration of an information processing device. FIG. 5 is a block diagram showing an example of the internal functional configuration of a learning device. FIG. 6 is a conceptual diagram showing an example of the contents of training data. FIG. 7 is a flowchart showing an example of the procedure of processing executed by a learning device. FIG. 8 is a flowchart showing the procedure of processing executed by an information processing device. FIG. 9 is a schematic diagram showing an example of an input image. FIG. 10 is a conceptual diagram showing a first example of the result of repeating the processes of S21 to S25. FIG. 11 is a conceptual diagram showing a second example of the result of repeating the processes of S21 to S25. FIG. 12 is a conceptual diagram showing a third example of the result of repeating the processes of S21 to S25. FIG. 13 is a conceptual diagram showing a fourth example of the result of repeating the processes of S21 to S25. FIG. 14 is a conceptual diagram showing a fifth example of the result of repeating the processes of S21 to S25.

[0009] The present disclosure will be described in detail below with reference to the drawings illustrating embodiments thereof. In this embodiment, a processing is performed to generate desired information from data related to a substrate processing apparatus using a base model. Figure 1 is a conceptual diagram showing an example configuration of the base model 1. The base model 1 includes an apparatus data encoder 11, a natural language encoder 12, an image encoder 13, and an integrated decoder 14.

[0010] The equipment data encoder 11 is a trained model that is trained to receive equipment data related to the substrate processing apparatus 21 and output a first feature value corresponding to the input equipment data. The equipment data encoder 11 performs a calculation to generate a first feature value that represents the feature of the equipment data. The equipment data encoder 11 corresponds to the first model. The equipment data is input to the equipment data encoder 11 from, for example, the substrate processing apparatus 21. The equipment data may also be input to the equipment data encoder 11 from a control device that controls the substrate processing apparatus 21.

[0011] The natural language encoder 12 is a trained model that is trained to receive a query expressed in a natural language and output a second feature value corresponding to the input query. The natural language encoder 12 performs a calculation to generate a second feature value that represents the features of the query. The natural language encoder 12 corresponds to the second model. The query is input, for example, by a user. The query requests some information from the base model 1.

[0012] The image encoder 13 is a trained model that is trained to receive an image related to substrate processing and output a third feature value corresponding to the input image. The image encoder 13 performs a calculation to generate a third feature value that represents a feature of the image. The image encoder 13 corresponds to the third model. The image related to substrate processing is, for example, an image that represents the shape of the substrate. The image related to substrate processing is, for example, generated by a measuring device 22 that measures the shape of the substrate, and input from the measuring device 22 to the image encoder 13.

[0013] The measuring device 22 is, for example, an imaging device that takes images of the substrate. Images taken by the measuring device 22, which is an imaging device, are input to the image encoder 13. Alternatively, the measuring device 22 may be a measuring instrument such as a sensor, and may be configured to generate images based on measured values, such as graphs that show changes in measured values ​​over time. Images based on measured values ​​measured by the measuring device 22, which is a measuring instrument, are input to the image encoder 13.

[0014] The integrated decoder 14 is a trained model trained to receive input of the first feature, the second feature, and the third feature and output an answer to a query. The integrated decoder 14 performs a calculation to generate an answer to a query according to the interrelationship between the first feature, the second feature, and the third feature. The integrated decoder 14 corresponds to an integrated model. The integrated decoder 14 is trained to output an answer even when no image is input to the image encoder 13 and no third feature is input to the integrated decoder 14. The integrated decoder 14 is also trained to output an answer even when no device data is input to the device data encoder 11 and no first feature is input to the integrated decoder 14.

[0015] The device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 are each configured using a neural network. For example, the device data encoder 11 or the image encoder 13 is configured using a convolutional neural network (CNN). The natural language encoder 12 may be configured using a language model such as BERT, GPT-4, Bard, or LLaMA. The integrated decoder 14 may be configured using GPT. A request for information regarding the substrate processing apparatus 21 is input to the board model 1 as a query, and the requested information regarding the substrate processing apparatus 21 is output from the board model 1 as a response to the query.

[0016] The equipment data input to the equipment data encoder 11 is data related to the substrate processing apparatus 21. The equipment data is, for example, data representing the state of the substrate processing apparatus 21. More specifically, the equipment data is, for example, sensor data obtained from a sensor provided in the substrate processing apparatus 21. The sensor data represents a time change in a physical quantity measured by a sensor, such as temperature, pressure, current, or voltage. The sensor data may represent a spatial distribution of the physical quantity. The sensor data may be data from an OES (Optical Emission Spectrometer). The equipment data may include multiple types of sensor data obtained from multiple sensors provided in the substrate processing apparatus 21. The sensor data numerically represents a change in the state of the substrate processing apparatus 21. Multiple types of sensor data represent a change in the state of the substrate processing apparatus 21 in more detail. Changes in the state of the substrate processing apparatus 21, such as temperature changes within the substrate processing apparatus 21, affect information about the substrate processing apparatus 21, such as the results of substrate processing. Therefore, the answer output from the board model 1 may change depending on the sensor data.

[0017] The equipment data may be log data showing the history of the state of the substrate processing apparatus 21. FIG. 2 is a diagram showing an example of the contents of the log data. Time and events occurring in the substrate processing apparatus 21 at each time are recorded in a table format. As shown in FIG. 2, the log data records the history of the state of the substrate processing apparatus 21 over time, such as the history of the operation of the substrate processing apparatus 21, the history of maintenance performed on the substrate processing apparatus 21, or the history of problems occurring in the substrate processing apparatus 21. Changes in the state of the substrate processing apparatus 21, such as the operating status of the substrate processing apparatus 21 or whether maintenance has been performed, affect information about the substrate processing apparatus 21, such as the results of substrate processing. Therefore, the response output from the substrate model 1 may change depending on the log data. The log data does not have to be in a table format. Alternatively, the equipment data may be recipe data specifying the contents of substrate processing.

[0018] The image related to substrate processing input to the image encoder 13 is, for example, an image representing the shape of the substrate. More specifically, the image related to substrate processing is, for example, an image of the surface of the substrate before or after processing captured by a camera. The image related to substrate processing is, for example, an image representing the surface or cross section of the substrate before or after processing measured by a scanning electron microscope (SEM). The image related to substrate processing may be, for example, an image representing the shape of the substrate measured using laser light, radio waves, or ultrasound. The image related to substrate processing may be, for example, an image representing the distribution of components of the substrate obtained by spectral analysis such as optical emission spectroscopy or fluorescence analysis. Furthermore, the image related to substrate processing is not limited to an image representing the entire substrate. For example, the image related to substrate processing may be an image representing a portion of the substrate, such as a bevel portion of the substrate.

[0019] The image related to the substrate processing is not limited to an image of the substrate itself. The image related to the substrate processing may be an image showing the state of the substrate processing apparatus 21. For example, it may be an image capturing the entire interior of a process chamber included in the substrate processing apparatus 21, or a portion thereof. The image related to the substrate processing may be an image based on information relating to the state of the substrate or the substrate processing apparatus 21 before, after, or during processing. For example, an image having a wavelength axis and a time axis obtained from OES data may be used as the image related to the substrate processing. For example, an image generated based on the results of measuring the substrate with a reflectometer may be used. For example, an image generated based on measurements obtained by a sensor provided in the substrate processing apparatus 21 or the measuring device 22 may be used. The image generated based on the measurements may be, for example, a graph showing the change in the measured values ​​over time, or an image showing the results of performing a short-time Fourier transform on a time series of the measured values.

[0020] The answer output from the integrated decoder 14 is a response to the query input to the natural language encoder 12. The answer output from the integrated decoder 14 includes text, numerical data, tabular data, graphs, or images. For example, a query asking what data is needed to predict the results of substrate processing is input to the natural language encoder 12, and a response suggesting the data needed to predict the results of substrate processing is output from the integrated decoder 14. For example, a query requesting the results of substrate processing is input to the natural language encoder 12, sensor data or log data is input to the equipment data encoder 11, and an image of the substrate before processing is input to the image encoder 13. In this case, the equipment data encoder 11 outputs, for example, performance data indicating the performance of the substrate processing.

[0021] FIG. 3 is a chart showing an example of performance data. Measurement results of the surface shape of a processed substrate are recorded in table format. Numerical values ​​representing the location on the substrate where the measurement was performed, roughness, surface shape, etc., are recorded in association with the substrate or measurement ID. In the example shown in FIG. 3, IDs AAAAA and BBBBB, which indicate two substrates, are recorded. X and Y in the figure indicate the location on the substrate where the measurement was performed, and Defect, Roughness, and Thickness indicate the amount of defects, surface roughness, and thickness, respectively, at the measurement location. CD (Critical Dimension) indicates the width of the groove formed on the substrate, and Top CD, Middle CD, and Bottom CD indicate the CD at the top, center, and bottom of the groove, respectively. Performance data may also be in a format other than a table, such as an image or graph.

[0022] The base model 1 is realized by an information processing device. FIG. 4 is a block diagram showing an example of the internal configuration of the information processing device 3. The information processing device 3 executes the information processing method. The information processing device 3 is configured using a computer such as a personal computer or a server device. The information processing device 3 includes a calculation unit 31, a memory 32, a storage unit 33, a reading unit 34, an input unit 35, and a display unit 36. The calculation unit 31 is configured using, for example, a central processing unit (CPU), a graphics processing unit (GPU), or a multi-core CPU. The calculation unit 31 may also be configured using a quantum computer. The memory 32 stores temporary data generated in conjunction with calculations. The memory 32 is, for example, a random access memory (RAM). The storage unit 33 is non-volatile, for example, a hard disk or non-volatile semiconductor memory. The reading unit 34 reads information from a recording medium 30 such as an optical disk or a portable memory.

[0023] The input unit 35 accepts data input. The input unit 35 includes an operation unit that accepts input of information such as text by accepting operations from a user. The operation unit is, for example, a touch panel, a keyboard, or a pointing device. For example, a query is input via the operation unit. The input unit 35 may include an input interface that inputs apparatus data from the substrate processing apparatus 21 or a control device that controls the substrate processing apparatus 21, or image data from the measuring apparatus 22. Data may be input to the input unit 35 by a method other than input from the substrate processing apparatus 21, the control device, or the measuring apparatus 22. The display unit 36 ​​displays images. The display unit 36 ​​is, for example, a liquid crystal display or an EL display (electroluminescent display).

[0024] The calculation unit 31 causes the reading unit 34 to read the computer program 331 recorded on the recording medium 30, and stores the read computer program 331 in the storage unit 33. The calculation unit 31 executes processing to realize the functions of the information processing device 3 in accordance with the computer program 331. The computer program 331 may be a program product. The computer program 331 may be stored in the storage unit 33 in advance, or may be downloaded from outside the information processing device 3. In this case, the information processing device 3 does not need to be equipped with the reading unit 34.

[0025] The computer program 331 can be deployed to run on a single computer, or on multiple computers located at one site or distributed across multiple sites and interconnected by a communications network. That is, the information processing device 3 may be configured with multiple computers, and the computer program 331 may be executed on multiple computers connected via a communications network. The information processing device 3 may be configured using a cloud server.

[0026] The information processing device 3 includes a base model 1. The base model 1 includes a device data encoder 11, a natural language encoder 12, an image encoder 13, and an integrated decoder 14. The device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 are realized by the calculation unit 31 executing information processing in accordance with a computer program 331. As described above, the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 are each configured using a neural network.

[0027] Any of the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 may be configured using hardware. For example, any of the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 may be configured using hardware including a processor and a memory for storing necessary programs and data. Alternatively, any of the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 may be implemented using a quantum computer. Alternatively, any of the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 may be provided external to the information processing device 3, and the information processing device 3 may perform processing using the external device data encoder 11, the natural language encoder 12, the image encoder 13, or the integrated decoder 14. For example, any of the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 may be implemented using the cloud.

[0028] Of the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14, the natural language encoder 12 and the image encoder 13 are general-purpose trained models. In this embodiment, the base model 1 is trained without training the natural language encoder 12 and the image encoder 13. The base model 1 is trained by the training device 4.

[0029] FIG. 5 is a block diagram showing an example of the internal functional configuration of the learning device 4. The learning device 4 is a computer such as a server device or a personal computer. The learning device 4 includes a calculation unit 41, a memory 42, a storage unit 43, a reading unit 44, an operation unit 45, and a display unit 46. The calculation unit 41 is configured using, for example, a CPU, a GPU, or a multi-core CPU. The calculation unit 41 may also be configured using a quantum computer. The memory 42 stores temporary data generated in conjunction with calculations. The memory 42 is, for example, a RAM. The reading unit 44 reads information from a recording medium 40 such as an optical disk or a portable memory. The storage unit 43 is non-volatile, for example, a hard disk or a non-volatile semiconductor memory.

[0030] The operation unit 45 receives information input by receiving operations from the user. The operation unit 45 is, for example, a keyboard, a pointing device, or a touch panel. The display unit 46 displays images. The display unit 46 is, for example, a liquid crystal display or an EL display.

[0031] The computer program 431 can be deployed to run on a single computer, or on multiple computers located at one site or distributed across multiple sites and interconnected by a communications network. That is, the learning device 4 may be configured with multiple computers, and the computer program 431 may run on multiple computers connected via a communications network. The learning device 4 may be configured using a cloud server.

[0032] The learning device 4 executes part of the information processing method. More specifically, the learning device 4 performs a process of training the device data encoder 11 and the integrated decoder 14. The learning device 4 is equipped with a base model 1 including the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14. The device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 are all trained models. The base model 1 is realized by the calculation unit 41 executing processing in accordance with a computer program 431. Parameters necessary for the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14 to perform calculations are stored in, for example, a memory unit 43.

[0033] The storage unit 43 stores training data 432 for training the base model 1. FIG. 6 is a conceptual diagram showing an example of the contents of the training data 432. The training data records a query, device data, an image, and an answer to the query, each associated with the other. The query, device data, and image to be input to the base model 1 are associated with the answer to be output from the base model 1 when the query, device data, and image are input. A combination of a query, device data, image, and answer constitutes one data set, and multiple data sets are recorded in the training data 432.

[0034] The equipment data included in the training data 432 is, for example, data obtained from the actual substrate processing apparatus 21 or a control device that controls the actual substrate processing apparatus 21. The images are, for example, images actually obtained by the measuring apparatus 22. The equipment data or images may be processed equipment data or images actually obtained, or may be arbitrarily created. Among the multiple data sets, there may be a data set in which part or all of the query, equipment data, image, or answer content is blank.

[0035] The answers included in the training data 432 may be created based on the manual or specifications of the substrate processing apparatus 21. The answers may be arbitrarily created to be answers corresponding to the queries. Data included in the answers, such as performance data, may be obtained by actual substrate processing, may be processed data that has actually been obtained, or may be arbitrarily created. The training data 432 includes a large number of data sets in which queries, apparatus data, images, and answers related to not only a single substrate processing apparatus 21 but also many types of substrate processing apparatuses are associated. Furthermore, the training data 432 includes a large number of data sets for each type of substrate processing apparatus. The training data 432 may include data sets in which queries, apparatus data, images, and answers related to different types of substrate processing apparatuses are associated.

[0036] The learning device 4 learns the board model 1 using training data. Note that pre-learning may be performed on the board model 1 before learning using the training data. For example, a query requesting creation of a correct sentence from a partially masked sentence regarding the substrate processing apparatus 21 based on a manual or specifications of the substrate processing apparatus 21 is input to the board model 1, and learning is performed so that the correct sentence is output from the board model 1 as a response. For example, a query requesting device data and substrate processing results based on a report regarding substrate processing previously performed in the substrate processing apparatus 21 is input to the board model 1, and learning is performed so that the results of the substrate processing, such as performance data, are output from the board model 1 as a response. The pre-learning may be performed by the learning device 4, or a board model 1 on which pre-learning has already been performed may be deployed to the learning device 4.

[0037] 7 is a flowchart showing an example of the processing procedure executed by the learning device 4. Hereinafter, steps are abbreviated as S. The learning device 4 executes the following processing by the calculation unit 41 executing information processing in accordance with the computer program 431. The calculation unit 41 reads out training data 432 stored in the storage unit 43, and the learning device 4 acquires the training data 432 (S11). In S11, the learning device 4 may acquire the training data 432 by inputting the training data 432 from an external source.

[0038] The learning device 4 then uses the training data 432 to train the device data encoder 11 and the integrated decoder 14 together (S12). In S12, the calculation unit 41 inputs the query, device data, and image included in the training data 432 to the base model 1. More specifically, the calculation unit 41 inputs the query included in the training data 432 to the natural language encoder 12, and inputs the device data associated with the query in the training data 432 to the device data encoder 11. The calculation unit 41 also inputs the image associated with the query in the training data 432 to the image encoder 13. If the content of the device data or the image is blank, either nothing or predetermined data corresponding to the blank is input to the device data encoder 11 or the image encoder 13.

[0039] The device data encoder 11 performs a calculation in response to input of device data, outputs a first feature, and inputs the first feature to the integrated decoder 14. The natural language encoder 12 performs a calculation in response to input of a query, outputs a second feature, and inputs the second feature to the integrated decoder 14. The image encoder 13 performs a calculation in response to input of an image, outputs a third feature, and inputs the third feature to the integrated decoder 14. The integrated decoder 14 performs a calculation in response to input of the first feature, second feature, and third feature, and outputs an answer to the query. In other words, an answer is output from the base model 1.

[0040] The calculation unit 41 adjusts the calculation parameters of the device data encoder 11 and the integrated decoder 14 so as to reduce the error between the answer output by the base model 1 and the answer associated in the training data 432 with the query, device data, and image input to the base model 1. For example, the calculation unit 41 adjusts the parameters using an error backpropagation method. At this time, the calculation unit 41 fixes the parameters for the natural language encoder 12 and the image encoder 13. In this way, the device data encoder 11 and the integrated decoder 14 are trained without training the natural language encoder 12 and the image encoder 13. The calculation unit 41 performs machine learning of the device data encoder 11 and the integrated decoder 14 by repeating processing using multiple data sets recorded in the training data 432 and adjusting the calculation parameters of the device data encoder 11 and the integrated decoder 14.

[0041] The learning device 4 then trains the integrative decoder 14 using the training data 432 (S13). In S13, similar to S12, the calculation unit 41 inputs the query, device data, and image included in the training data 432 to the base model 1. The device data encoder 11, the natural language encoder 12, and the image encoder 13 output the first feature amount, the second feature amount, and the third feature amount, which are input to the integrative decoder 14, and the integrative decoder 14 outputs an answer.

[0042] The calculation unit 41 adjusts the calculation parameters of the integrating decoder 14 so as to reduce the error between the answer output by the base model 1 and the answer associated in the training data 432 with the query, device data, and image input to the base model 1. At this time, the calculation unit 41 fixes the parameters for the device data encoder 11, the natural language encoder 12, and the image encoder 13. In this way, only the integrating decoder 14 is trained. The calculation unit 41 performs machine learning of the integrating decoder 14 by repeating processing using multiple data sets recorded in the training data 432 and adjusting the calculation parameters of the integrating decoder 14.

[0043] Alternatively, the integrating decoder 14 may be trained by learning using human feedback. The calculation unit 41 acquires the answer output by the base model 1 and displays it on the display unit 46. The user checks the displayed answer, determines the difference between the displayed answer and the appropriate answer to be output, and inputs information indicating the determined difference to the learning device 4 by operating the operation unit 45. The calculation unit 41 adjusts the calculation parameters of the integrating decoder 14 so as to reduce the error corresponding to the input difference. The query, device data, or image to be input to the base model 1 may be input to the learning device 4 by the user operating the operation unit 45.

[0044] In S12 or S13, the calculation unit 41 may input a query requesting reconstruction of equipment data or images to the substrate model 1, and perform learning so that the substrate model 1 outputs reconstructed equipment data or images. Since learning is performed using a large number of data sets related to various types of substrate processing apparatuses recorded in the training data 432, the substrate model 1 becomes a model adapted to various types of substrate processing apparatuses. Furthermore, learning is performed using a large number of data sets related to each type of substrate processing apparatus. In this way, the substrate model 1 is trained to adapt to various settings, situations, and environments related to various substrate processing apparatuses, and is able to provide appropriate output in response to a variety of inputs. The calculation unit 41 stores the final parameters adjusted in S12 and S13 in the memory unit 43. After S13 is completed, the learning device 4 terminates processing.

[0045] The base model 1, in which the equipment data encoder 11 and the integrated decoder 14 have been trained through the processes of S11 to S13, is deployed to the information processing device 3. For example, the final parameters of the base model 1 adjusted through the processes of S11 to S13 are input to the information processing device 3 through the input unit 35 and stored in the memory unit 33. The base model 1 is realized by the calculation unit 31 executing information processing using the stored parameters. Note that the information processing device 3 may also have the function of the learning device 4. In other words, the information processing device 3 may execute the processes of S11 to S13. The base model 1, in which the equipment data encoder 11 and the integrated decoder 14 have been trained based on data relating to a variety of substrate processing devices, is capable of inputting and outputting information relating to a variety of substrate processing devices.

[0046] The information processing device 3 uses the substrate model 1 to perform a process of inferring various pieces of information related to the substrate processing device 21. FIG. 8 is a flowchart showing the procedure of the process executed by the information processing device 3. The calculation unit 31 executes information processing in accordance with the computer program 331, causing the information processing device 3 to perform the following process. The information processing device 3 inputs device data, a query, and an image to the substrate model 1 (S21). In S21, the device data is input to the input unit 35 from the substrate processing device 21 or the control device, the query is input by the user operating an operation unit included in the input unit 35, and the image is input to the input unit 35 from the measuring device 22. The information processing device 3 inputs the input query and a prompt including the selected device data and image to the substrate model 1.

[0047] For example, the information processing device 3 displays an input image on the display unit 36 ​​and uses the input image to input device data, a query, and an image. FIG. 9 is a schematic diagram showing an example of an input image. The input image includes an input field 51 for inputting a query. The query is input in text format into the input field 51 by a user operating an operation unit included in the input unit 35. The input image also includes selection icons 52 and 53 for selecting device data and an image. When the selection icon 52 or 53 in the input image is designated by a user operating the operation unit, multiple device data or images stored in the storage unit 33 are displayed, and the device data or image to be input is selected. FIG. 9 shows an example in which "log data 01" is selected as device data, and "image 02" is selected from multiple images.

[0048] The calculation unit 31 generates an input image including an input field 51 and selection icons 52 and 53, and displays the image on the display unit 36. The positional relationship between the input field 51 and the selection icons 52 and 53 shown in FIG. 9 is an example, and the input field 51 and the selection icons 52 and 53 may be located in other positions. The user interface for inputting device data or images may be in a form other than the selection icons 52 or 53. The calculation unit 31 displays multiple input images on the display unit 36, and the input field 51 and the user interface for inputting device data or images may be included in separate input images. The calculation unit 31 accepts device data, queries, and images input by a user operating the operation unit using the input images.

[0049] In S21, the calculation unit 31 inputs the equipment data to the equipment data encoder 11, inputs the query to the natural language encoder 12, and inputs the image to the image encoder 13. The equipment data or the image may be blank. The calculation unit 31 may input a prompt including content referencing the equipment data or the image in addition to the query to the natural language encoder 12. For example, the calculation unit 31 may generate a prompt including information specifying the equipment data or the image, such as "Please generate performance data by referring to the following log data and images before and after substrate processing. {logdata01.csv}, {image01.jpeg}, {image01.jpeg}," and input the prompt to the natural language encoder 12. At this time, the calculation unit 31 inputs the equipment data or the image referenced in the prompt to the equipment data encoder 11 or the image encoder 13.

[0050] After the device data, query, and image are input to the base model 1, the device data encoder 11 performs a calculation and outputs a first feature value corresponding to the input device data. The natural language encoder 12 performs a calculation and outputs a second feature value corresponding to the input query. The image encoder 13 performs a calculation and outputs a third feature value corresponding to the input image.

[0051] The information processing device 3 then acquires a first feature amount, a second feature amount, and a third feature amount (S22). In S22, the calculation unit 31 acquires the first feature amount, the second feature amount, and the third feature amount output by the device data encoder 11, the natural language encoder 12, and the image encoder 13. The information processing device 3 inputs the acquired first feature amount, the second feature amount, and the third feature amount to the integrated decoder 14 (S23). The integrated decoder 14 performs calculations and outputs a response corresponding to the input first feature amount, the second feature amount, and the third feature amount.

[0052] The information processing device 3 acquires the answer to the query output by the base model 1 (S24). In S24, the calculation unit 31 acquires the answer to the query output by the integrated decoder 14, and the information processing device 3 outputs the acquired answer (S25). In S25, the calculation unit 31 displays the answer to the query on the display unit 36. The calculation unit 31 stores the acquired answer in the memory unit 33. After S25 is completed, the information processing device 3 ends the processing.

[0053] The information processing device 3 executes the processes of S21 to S25 as needed. For example, the information processing device 3 repeats the processes of S21 to S25 in response to an operation from the user. FIG. 10 is a conceptual diagram showing a first example of the results of repeating the processes of S21 to S25. FIG. 10 shows an example of the contents of a query, etc. input to the board model 1 and a response output from the board model 1. A query inquiring about a method for estimating the performance of substrate processing is input, and a response specifying the necessary data is output. A query, log data, which is equipment data, and an image are input, and performance data corresponding to the log data and image is output as a response.

[0054] 10, a prompt including a request for advice to improve the quality of an answer such as performance data in addition to a query may be input to the base model 1. As a result of repeating the process in accordance with the advice, an answer such as performance data of improved quality is output. The calculation unit 31 can display performance data such as that shown in FIG. 3 on the display unit 36.

[0055] 11 is a conceptual diagram showing a second example of the results of repeating the processes of S21 to S25. A query requesting a proposal of recipe data suitable for a substrate is input, and a response specifying the sensor data of the substrate processing apparatus 21 and an image of the substrate before processing as the necessary data is output. The query, the sensor data, which is apparatus data, and the image of the substrate before processing are input, and recipe data is output as the response.

[0056] 12 is a conceptual diagram showing a third example of the results of repeating the processes of S21 to S25. A query requesting a prediction of the maintenance timing for the substrate processing apparatus 21 is input, and a response specifying log data of the substrate processing apparatus 21 and an image inside the substrate processing apparatus 21 as the required data is output. For example, the image inside the substrate processing apparatus 21 is an image captured of the inside of a process chamber of the substrate processing apparatus 21. The query, the log data, which is apparatus data, and the image inside the substrate processing apparatus 21 are input, and a result predicting the maintenance timing for the substrate processing apparatus 21 is output as a response.

[0057] 13 is a conceptual diagram showing a fourth example of the results of repeating the processes of S21 to S25. A query inquiring about a method for increasing the operating rate of the substrate processing apparatus 21 is input, and a response specifying the log data of the substrate processing apparatus 21 as the required data is output. A query requesting a setting proposal for increasing the operating rate of the substrate processing apparatus 21 and the log data, which is apparatus data, are input, and effective settings for improving the operating rate of the substrate processing apparatus 21 are output as a response.

[0058] FIG. 14 is a conceptual diagram showing a fifth example of the results of repeating the processes of S21 to S25. A query inquiring about how to deal with an alarm in the substrate processing apparatus 21 is input, and a response requesting input of the alarm content is output. The alarms are various alarms output by the substrate processing apparatus 21. After specifying the content of the alarm, a query inquiring about how to deal with the alarm is input, and the response to the alarm is output. As shown in FIGS. 10 to 14 , input of queries and the like to the substrate model 1 and output from the substrate model 1 are appropriately repeated. Depending on the query, apparatus data, or image input to the substrate model 1, the response output from the substrate model 1 can take various forms, such as a character string, a data file, or an image. By adjusting the input to the substrate model 1, the user can infer various information about the substrate processing apparatus.

[0059] As described above in detail, device data is input to the device data encoder 11, which outputs a first feature, a query is input to the natural language encoder 12, which outputs a second feature, and an image is input to the image encoder 13, which outputs a third feature. The first feature, the second feature, and the third feature are input to the integrated decoder 14, which outputs an answer to the query. A base model 1 is obtained that includes the device data encoder 11, the natural language encoder 12, the image encoder 13, and the integrated decoder 14. By using the natural language encoder 12 and the image encoder 13 as general-purpose trained models and training the device data encoder 11 and the integrated decoder 14 so that they can be applied to various substrate processing apparatuses, a base model 1 that can infer information about various substrate processing apparatuses can be realized.

[0060] The trained substrate model 1 outputs a response in response to an input query. Information about the substrate processing apparatus is inferred in response to inputting apparatus data and images related to the substrate processing apparatus and a query requesting inference of information about the substrate processing apparatus to the substrate model 1. For example, the quality of the processed substrate can be inferred.

[0061] The board model 1 is adapted to a wide variety of substrate processing apparatuses by being trained using a large number of data sets relating to a wide variety of substrate processing apparatuses. Furthermore, by being trained using a large number of data sets relating to each type of substrate processing apparatus, the board model 1 can provide appropriate output in response to various inputs relating to each substrate processing apparatus. Therefore, the board model 1 has high versatility, enabling appropriate output in response to various inputs relating to various types of substrate processing apparatuses. By using this board model 1, various pieces of information relating to various substrate processing apparatuses can be estimated. Because the board model 1 is highly versatile, there is no need to develop a board model for each individual substrate processing apparatus or to tune the board model for each individual substrate processing apparatus. This reduces the amount of work required to develop or operate the board model, making it possible to easily realize the board model 1 that estimates information relating to the substrate processing apparatus 21.

[0062] In this embodiment, an example has been shown in which the first model is the device data encoder 11, the second model is the natural language encoder 12, the third model is the image encoder 13, and the integrated model is the integrated decoder 14. The first model, the second model, the third model, and the integrated model may be configured by a combination of learning models other than the combination in which the first model, the second model, and the third model are encoders and the integrated model is a decoder.

[0063] The present invention is not limited to the contents of the above-described embodiment, and various modifications are possible within the scope of the claims. In other words, embodiments obtained by combining technical means modified appropriately within the scope of the claims are also included in the technical scope of the present invention.

[0064] The matters described in each embodiment can be combined with each other. Furthermore, the independent claims and dependent claims described in the claims can be combined with each other in any and all combinations, regardless of the reference format. Furthermore, the claims do not use a format in which a claim references two or more other claims (multiple claim format), but this is not limited to this. They may be written using a multiple claim format or a format in which multiple claims (multi-multi claim) reference at least one other multiple claim.

[0065] REFERENCE SIGNS LIST 1 Base model 11 Device data encoder (first model) 12 Natural language encoder (second model) 13 Image encoder (third model) 14 Integrated decoder (integrated model) 21 Substrate processing apparatus 3 Information processing apparatus 30 Recording medium 31 Calculation unit 33 Storage unit 331 Computer program 4 Learning apparatus 41 Calculation unit 431 Computer program

Claims

1. An information processing method comprising: inputting data relating to a substrate processing apparatus into a first model; acquiring a first feature output by the first model; inputting a query relating to the substrate processing apparatus into a second model; acquiring a second feature output by the second model; inputting the first feature and the second feature into an integrated model; acquiring an answer to the query output by the integrated model; and outputting the acquired answer.

2. The information processing method according to claim 1, further comprising the steps of: inputting an image relating to substrate processing into a third model; obtaining a third feature quantity output by the third model; inputting the first feature quantity, the second feature quantity and the third feature quantity into the integrated model; and obtaining the answer output by the integrated model.

3. The information processing method according to claim 2, wherein the first model, the second model and the third model are encoders, and the integrated model is a decoder.

4. The information processing method of claim 1, further comprising: inputting a query inquiring about the results of substrate processing into the second model as the query; inputting sensor data obtained from a sensor provided in the substrate processing apparatus or log data showing the history of the state of the substrate processing apparatus into the first model as data relating to the substrate processing apparatus; and obtaining performance data showing the results of the substrate processing output by the integrated model as a response to the query.

5. The information processing method of claim 2, further comprising: inputting, as the query, a query inquiring about the results of the substrate processing to the second model; inputting, as the data relating to the substrate processing apparatus, sensor data obtained from a sensor provided in the substrate processing apparatus or log data showing the history of the state of the substrate processing apparatus to the first model; inputting, as the image relating to the substrate processing, an image showing the substrate before or after processing to the third model; and obtaining, as a response to the query, performance data showing the results of the substrate processing output by the integrated model.

6. The information processing method of claim 1, wherein the query is input to the second model as a query requesting a proposal for recipe data, a query requesting a prediction of a maintenance timing for the substrate processing apparatus, a query inquiring about a method for increasing the operating rate of the substrate processing apparatus, or a query inquiring about a method for dealing with an alarm of the substrate processing apparatus.

7. The information processing method according to claim 1, further comprising the step of inputting a plurality of types of sensor data obtained from a plurality of sensors provided in the substrate processing apparatus into the first model as the data relating to the substrate processing apparatus.

8. The information processing method of claim 1, further comprising: acquiring training data including data relating to a substrate processing apparatus, a query relating to the substrate processing apparatus, and an answer to the query; using the training data to train the first model and the integrated model together; and then training the integrated model using the training data.

9. An information processing device comprising a calculation unit that inputs data related to a substrate processing apparatus into a first model, and acquires a first feature output by the first model, inputs a query related to the substrate processing apparatus into a second model, and acquires a second feature output by the second model, inputs the first feature and the second feature into an integrated model, and acquires an answer to the query output by the integrated model, and outputs the acquired answer.

10. A computer program that causes a computer to execute the following processes: inputting data related to a substrate processing apparatus into a first model, acquiring a first feature output by the first model; inputting a query related to the substrate processing apparatus into a second model, acquiring a second feature output by the second model; inputting the first feature and the second feature into an integrated model, acquiring an answer to the query output by the integrated model; and outputting the acquired answer.