Information processing method, information processing apparatus, and substrate processing system

Abstract

Provided are an information processing method, an information processing device, and a substrate processing system.　The present invention comprises: a step for acquiring time-series data from a sensor which is provided in a substrate processing apparatus; a step for training, on the basis of the acquired time-series data, a first learning model that outputs information pertaining to the substrate processing apparatus when having received input of the time-series data from the sensor; and a step for inputting time-series data from the sensor into the first learning model after the training, and outputting an estimation result based on information obtained from the first learning model.

Description

Technical Field

[0001] The present invention relates to an information processing method, an information processing apparatus, and a substrate processing system.

Background Art

[0002] In a substrate processing system, a plurality of edge devices are used, and various processes are executed in a plurality of types of chambers having the plurality of edge devices.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Patent Document 2

Summary of the Invention

Problems to be Solved by the Invention

[0004] In view of such circumstances, the present disclosure provides an information processing method, an information processing apparatus, and a substrate processing system that can efficiently create and manage a model applied to various edge devices provided in a substrate processing system.

Means for Solving the Problems

[0005] An information processing method according to one embodiment of the present invention includes a step of acquiring time-series data from a sensor provided in a substrate processing apparatus, a step of training a first learning model that outputs information regarding the substrate processing apparatus when the time-series data from the sensor is input based on the acquired time-series data, and a step of inputting the time-series data from the sensor to the trained first learning model and outputting an estimation result based on the information obtained from the first learning model.

Effects of the Invention

[0006] According to this disclosure, inference models for various edge devices installed in a substrate processing system can be efficiently created and managed. [Brief explanation of the drawing]

[0007] [Figure 1] This figure shows an example configuration of a substrate processing system according to an embodiment. [Figure 2] This is a schematic diagram showing the configuration of the drive system of a substrate processing device. [Figure 3] This is a cross-sectional view showing an example of the chamber configuration. [Figure 4] This is a block diagram illustrating the configuration of the control system for a substrate processing device. [Figure 5] This is a block diagram showing the internal configuration of an edge device provided by a substrate processing apparatus. [Figure 6] This is a schematic diagram illustrating an example of the configuration of an observation model provided by an edge device. [Figure 7] This is a schematic diagram illustrating an example of the configuration of a control model for an edge device. [Figure 8] This is a block diagram showing the internal configuration of a control device provided by a substrate processing apparatus. [Figure 9] This is a conceptual diagram showing an example of a database configuration. [Figure 10] This is a block diagram showing the internal configuration of the server group. [Figure 11] This is a conceptual diagram showing an example of a database configuration. [Figure 12] This is a flowchart showing the procedure for generating the first learned model using an edge device. [Figure 13] This flowchart shows the procedures for processing performed inside the substrate processing unit during the operational phase. [Figure 14] This flowchart shows the procedure for processing performed between the substrate processing equipment and the equipment group server during the operational phase. [Figure 15] This is a schematic diagram showing an example of how evaluation results are displayed. [Modes for carrying out the invention]

[0008] Hereinafter, an embodiment will be described with reference to the drawings. In the description, elements having the same element or the same function are denoted by the same reference numerals, and redundant descriptions are omitted.

[0009] (Embodiment 1) FIG. 1 is a diagram showing a configuration example of a substrate processing system according to an embodiment. The substrate processing system according to the embodiment is a system for managing a learning model applied to a plurality of substrate processing apparatuses 100A to 100D. The substrate processing system includes a plurality of substrate processing apparatuses 100A to 100D and a device group server 200 that collects data from the plurality of substrate processing apparatuses 100A to 100D. Each of the substrate processing apparatuses 100A to 100D and the device group server 200 are communicably connected via a communication network NW such as a LAN (Local Area Network) or a dedicated line.

[0010] In the present embodiment, the substrate processing system is configured to include four substrate processing apparatuses 100A to 100D, but the number of apparatuses is not limited to four. In the following description, when it is not necessary to distinguish between the substrate processing apparatuses 100A to 100D, they may also be described as the substrate processing apparatus 100 (see FIG. 2). In the present embodiment, the device group server 200 may be a single computer, or may be a computer system configured by a plurality of computers, peripheral devices, or the like. Further, the device group server 200 may be a virtual machine in which the entity is virtualized, or may be a cloud.

[0011] FIG. 2 is a schematic diagram showing a configuration of a drive system of the substrate processing apparatus 100. The substrate processing apparatus 100 includes a transfer unit HU for loading and unloading the substrate W, and a processing unit SU for performing substrate processing on the substrate W.

[0012] <0OO0090>The transfer unit HU includes a cassette stage 10 and a transfer stage 20. The cassette stage 10 includes a cassette container 11. In the cassette container 11, for example, up to 25 substrates W are accommodated in a stacked state.

[0013] The transfer stage 20 includes a substrate transfer mechanism 21 for transferring the substrate W. The substrate transfer mechanism 21 has two transfer arms 21A and 21B that hold the substrate W substantially horizontally. The substrate transfer mechanism 21 uses the transfer arms 21A and 21B to take out the substrate W from the cassette container 11 one by one. The substrate transfer mechanism 21 transfers the substrate W taken out from the cassette container 11 to either one of the load lock chambers 25A and 25B. The load lock chambers 25A and 25B connect the transfer unit HU to the processing unit SU.

[0014] The processing unit SU includes a transfer chamber 30 and four chambers 40A to 40D. The transfer chamber 30 has a sealable structure formed, for example, in a polygonal shape (hexagonal shape in the illustrated example) when viewed from above. The transfer chamber 30 is connected to each of the chambers 40A to 40D via a gate valve that can be sealed airtight. The transfer chamber 30 includes a substrate transfer mechanism 31 for transferring the substrate W. The substrate transfer mechanism 31 has two transfer arms 31A and 31B that hold the substrate W substantially horizontally. The substrate transfer mechanism 31 uses the transfer arms 31A and 31B to take out the substrate W from the load lock chambers 25A and 25B and transfer the taken-out substrate W to any one of the chambers 40A to 40D.

[0015] With this configuration, the processing unit SU transports the substrates W, which have been conveyed to the load lock chambers 25A and 25B, to the chambers 40A to 40D via the transfer chamber 30, and performs substrate processing in the chambers 40A to 40D. After performing the substrate processing, the processing unit SU removes the processed substrates W from the chambers 40A to 40D and transports the removed substrates W back to the load lock chambers 25A and 25B via the transfer chamber 30. An example of substrate processing performed by chambers 40A to 40D is film deposition processing by CVD (Chemical Vapor Deposition). Alternatively, the substrate processing performed by chambers 40A to 40D may be diffusion processing, etching processing, ashing processing, sputtering processing, etc. Furthermore, although the example in Figure 2 shows a single-wafer substrate processing apparatus 100 that processes substrates W one by one by taking them out of the cassette container 11, the substrate processing apparatus 100 may also be a batch-type substrate processing apparatus that processes multiple substrates W simultaneously, and the transport unit HU can adopt any configuration.

[0016] In the following explanation, if it is not necessary to distinguish between chambers 40A to 40D, they will simply be referred to as chamber 40 (see Figure 3).

[0017] Figure 3 is a cross-sectional view showing an example configuration of the chamber 40. The chamber 40 shown as an example in Figure 3 is a device for performing a film deposition process on a substrate W, and comprises a processing chamber 41 in which the substrate processing is carried out, and an exhaust chamber 42 communicating with the processing chamber 41.

[0018] The processing chamber 41 has a plate-shaped ceiling portion 411 and a bottom portion 413, as well as a side wall portion 412 connecting the ceiling portion 411 and the bottom portion 413. The processing chamber 41 is, for example, roughly cylindrical in shape. An inlet / outlet is formed in the side wall portion 412 for loading and unloading substrates W to and from the transfer chamber 30. By opening the gate valve provided between the processing chamber 41 and the transfer chamber 30, it becomes possible to load and unload substrates W through this inlet / outlet. An opening 413a is formed in the center of the bottom portion 413. The exhaust chamber 42 is connected to the bottom portion 413 of the processing chamber 41 so as to communicate with the opening 413a.

[0019] The exhaust chamber 42 has an annular flange portion 421, a plate-shaped bottom portion 423, and a side wall portion 422 connecting the flange portion 421 and the bottom portion 423. The flange portion 421 is joined to the bottom portion 413 of the processing chamber 41. An exhaust hole 424 is formed in the side wall portion 422.

[0020] The processing chamber 41 and the exhaust chamber 42 are configured to maintain a reduced pressure atmosphere (vacuum) within their internal spaces. O-rings are interposed as sealing members at the joints between the processing chamber 41 and the exhaust chamber 42, as well as at the joints between the components constituting the processing chamber 41 and the exhaust chamber 42, in order to ensure airtightness of the joints.

[0021] The chamber 40 includes an exhaust device 51 located outside the processing chamber 41 and the exhaust chamber 42, an exhaust pipe 52 connecting the exhaust port 424 to the exhaust device 51, and a valve 53 provided in the middle of the exhaust pipe 52. When the valve 53 is closed, it maintains the airtightness of the processing chamber 41 and the exhaust chamber 42, and when it is open, it allows the exhaust device 51 to reduce the pressure in the processing chamber 41 and the exhaust chamber 42. By operating the exhaust device 51, the internal space of the processing chamber 41 and the exhaust chamber 42 is reduced to the required vacuum level.

[0022] The chamber 40 comprises a susceptor 61 positioned within the processing chamber 41, and a support member 62 that supports the susceptor 61 within the processing chamber 41 and the exhaust chamber 42. The susceptor 61 is a substrate mounting platform for horizontally supporting the substrate W. The susceptor 61 has a substrate mounting surface (upper surface) on which the substrate W is placed, and a lower surface on the opposite side. One end of the support member 62 is fixed to the center of the lower surface of the susceptor 61. The other end of the support member 62 is fixed to the bottom 423 of the exhaust chamber 42.

[0023] Although not shown in the figures, the susceptor 61 has a plurality of support pins that are provided to protrude from and retract relative to the substrate mounting surface. The plurality of support pins are configured to be displaced vertically by an arbitrary lifting mechanism, and in the raised position, the substrate W can be transferred between the susceptor and the substrate transport mechanism 31.

[0024] Chamber 40 comprises a heater 63, a heater power supply 64, and a temperature sensor TS. The heater 63 and the temperature-measuring portion of the temperature sensor TS are embedded in the susceptor 61. The heater power supply 64 is located outside the processing chamber 41 and the exhaust chamber 42. The heater 63 is connected to the heater power supply 64, for example, via wiring that runs through the inside of a support member 62. The heater power supply 64 supplies an electrical output to the heater 63 to heat the substrate W placed on the susceptor 61 to a desired temperature. The temperature of the susceptor 61 is measured by the temperature sensor TS. The temperature sensor TS is constructed using known components such as a thermocouple or a thermistor.

[0025] Chamber 40 is equipped with a shower head 71 located on the ceiling 411 of the processing chamber 41. The shower head 71 has a gas diffusion space 71a formed inside it and a plurality of gas discharge holes 71b formed to penetrate from the gas diffusion space 71a toward the susceptor 61.

[0026] Chamber 40 is located on the opposite side of the showerhead 71 from the multiple gas discharge holes 71b and includes a gas introduction pipe 72 that communicates with the gas diffusion space 71a, a gas supply source 73 located outside the processing chamber 41 and exhaust chamber 42, a gas piping 74 connecting the gas introduction pipe 72 and the gas supply source 73, and an MFC (mass flow controller) 75 and a valve (not shown) located in the middle of the gas piping 74. The gas supply source 73 supplies the showerhead 71 with film-forming raw material gas used in the film-forming process, cleaning gas for cleaning the processing chamber 41 and exhaust chamber 42, and purge gas for replacing the atmosphere in the processing chamber 41 and exhaust chamber 42. These gases are supplied to the gas diffusion space 71a via the gas piping 74 and the gas introduction pipe 72 and discharged into the processing chamber 41 from the multiple gas discharge holes 71b.

[0027] Chamber 40 includes a high-frequency power supply 76 located outside the processing chamber 41 and exhaust chamber 42, wiring 77 connecting the shower head 71 and the high-frequency power supply 76, and a matching unit 78 provided in the middle of the wiring 77. The high-frequency power supply 76 supplies high-frequency power to the shower head 71 to plasmaize the film-forming raw material gas supplied into the processing chamber 41.

[0028] With the above configuration, the chamber 40 can perform a film deposition process on a substrate W placed on a susceptor 61. Specifically, the substrate W to be processed is transported into the vacuum chamber 41, the substrate W placed on the susceptor 61 is heated by the heater 63, and a raw material gas is supplied to the substrate W from the shower head 71, thereby forming a thin film on the surface of the substrate W. In addition, high-frequency power may be supplied to the shower head 71 from the high-frequency power supply 76 to accelerate the film deposition reaction. In this case, it becomes possible to plasmaize the raw material gas supplied into the chamber 41 via the shower head 71 and deposit a film.

[0029] In the example shown in Figure 3, a chamber 40 for performing a film deposition process on the substrate W was described, but the chamber 40 may also be a process module for performing diffusion, etching, ashing, sputtering, and other processes on the substrate W.

[0030] Figure 4 is a block diagram illustrating the configuration of the control system of the substrate processing apparatus 100. The substrate processing apparatus 100 includes various sensors S1 to S3, edge devices 110 to 130 to which the data output from sensors S1 to S3 is input, actuators A1 to A3 controlled by the edge devices 110 to 130, and a control device 150 that controls the operation of the entire apparatus.

[0031] Sensors S1 to S3 are installed on the substrate processing device 100 and measure the physical quantity to be measured in a time series. Sensors S1 to S3 output time series data (hereinafter also referred to as sensor data) showing the measurement results to the subsequent edge devices 110 to 130.

[0032] An example of sensor S1 is an RF (Radio Frequency) sensor. The RF sensor is installed on the output side of the high-frequency power supply 76 and measures the RF power of the high-frequency power supply 76 in a time series. Alternatively, the RF sensor may be a sensor that measures voltage, current, capacitance, impedance, phase, load power, etc., in a time series. An example of sensor S2 is a temperature sensor. The temperature sensor includes a temperature sensor TS in which the temperature measuring part is embedded in the susceptor 61 and measures the temperature of the substrate mounting surface (i.e., the temperature of the substrate W to be processed) in a time series. Alternatively, the temperature sensor may be a sensor that measures electrode temperature, the internal temperature of the processing chamber 41, etc., in a time series. An example of sensor S3 is a torque sensor. The torque sensor measures the torque received by actuators (e.g., actuator A3) mounted on the substrate transport mechanisms 21, 31 in a time series.

[0033] The sensors S1 to S3 mounted on the substrate processing apparatus 100 are not limited to the RF sensor, temperature sensor, and torque sensor described above. Furthermore, the number of sensors mounted on the substrate processing apparatus 100 is not limited to three. For example, the substrate processing apparatus 100 may include one or more sensors, including a gas sensor, an OES (Optical Emission Spectroscopy) sensor, and a flow rate sensor. Here, the gas sensor is installed in the processing chamber 41 and measures the amount of a specific component of the gas filling the processing chamber 41 over time. Examples of gas sensors include mass spectrometers, infrared spectrometers, and gas chromatographs. The OES sensor is installed in the processing chamber 41 and measures the plasma emission intensity inside the processing chamber 41 over time. The flow rate sensor is installed in the processing chamber 41 and measures the flow rate of the gas introduced into the processing chamber 41 over time.

[0034] The sampling rates for sensors S1 to S3 are arbitrary and can be set appropriately for each sensor. For example, the sampling rate for the RF sensor is 1 to 10 μsec. The sampling rate for the temperature sensor is, for example, 100 msec. The sampling rate for the torque sensor is, for example, 2.5 msec. The sampling rate for the OES sensor is, for example, 10 to 100 msec. The sampling rate for the gas sensor is, for example, 1 to 10 msec. The sampling rate for the flow sensor is, for example, 10 msec.

[0035] The edge devices 110 to 130 perform processes such as estimating the state of the substrate processing device 100 and estimating control values ​​for actuators A1 to A3 based on sensor data input from sensors S1 to S3. For example, the edge devices 110 to 130 output the state estimation result to the control device 150 and control the operation of actuators A1 to A3 based on the control value estimation result. The internal configuration of the edge devices 110 to 130 and the content of the processes performed by edge device 110 will be described in detail later.

[0036] Actuators A1 to A3 are controlled by edge devices 110 to 130. In this embodiment, the term "actuator" is used to refer to various drive circuits, including electrical circuits, not limited to mechanical elements that convert electrical energy into physical momentum, such as motors.

[0037] For example, if sensor S1 is an RF sensor, actuator A1 may be a high-frequency power supply 76. In this case, edge device 110 acquires time-series data of RF power from the RF sensor, estimates a control value for the high-frequency power supply 76 based on the acquired time-series data, and controls the operation of the high-frequency power supply 76 based on the estimated control value. Also, if sensor S2 is a temperature sensor TS, actuator A2 may be a heater power supply 64. In this case, edge device 120 acquires time-series data of temperature from the temperature sensor TS, estimates a control value for the heater power supply 64 based on the acquired time-series data, and controls the operation of the heater power supply 64 based on the estimated control value. Furthermore, if sensor S3 is a torque sensor, actuator A3 may be a motor provided in the substrate transport mechanisms 21 and 31. In this case, edge device 130 acquires time-series data of torque received by the motor drive shaft from the torque sensor, estimates a control value for the substrate transport mechanisms 21 and 31 based on the acquired time-series data, and controls the operation of the substrate transport mechanisms 21 and 31 based on the estimated control value.

[0038] In this embodiment, one sensor and one actuator are connected to each of the edge devices 110 to 130, but the number of sensors and actuators connected to each of the edge devices 110 to 130 is not limited to one. Multiple sensors and multiple actuators may be connected to each of the edge devices 110 to 130. Also, in this embodiment, the substrate processing apparatus 100 is configured to have three edge devices 110 to 130, but the number of edge devices mounted on the substrate processing apparatus 100 is not limited to three; it may have one or more edge devices.

[0039] The control device 150 controls the overall operation of the substrate processing apparatus 100 based on various information input from the edge devices 110-130 and various information input from external sources. The internal configuration of the control device 150 and the content of the processing performed by the control device 150 will be described in detail later.

[0040] Figure 5 is a block diagram showing the internal configuration of the edge device 110 provided in the substrate processing apparatus 100. The edge device 110 is a dedicated or general-purpose computer installed in the substrate processing apparatus 100, and includes a control unit 111, a storage unit 112, an input unit 113, an output unit 114, a communication unit 115, and the like. Based on sensor data from sensor S1 input through the input unit 113, the edge device 110 monitors the state of the substrate processing apparatus 100 and controls the operation of actuator A1.

[0041] The control unit 111 includes a CPU (Central Processing Unit), ROM (Read Only Memory), RAM (Random Access Memory), and the like. The ROM in the control unit 111 stores control programs that control the operation of each hardware component of the edge device 110. The CPU in the control unit 111 reads and executes the control programs stored in the ROM and various computer programs stored in the memory unit 112, and controls the operation of each hardware component, thereby making the entire device function as an information processing device according to this disclosure. The RAM in the control unit 111 temporarily stores data used during the execution of calculations.

[0042] In this embodiment, the control unit 111 is configured to include a CPU, ROM, and RAM, but the configuration of the control unit 111 is not limited to the above. The control unit 111 may be one or more control circuits or arithmetic circuits that include, for example, a GPU (Graphics Processing Unit), FPGA (Field Programmable Gate Array), DSP (Digital Signal Processor), quantum processor, volatile or non-volatile memory, etc. Furthermore, the control unit 111 may include functions such as a clock that outputs date and time information, a timer that measures the elapsed time from the time a measurement start instruction is given to the time a measurement end instruction is given, and a counter that counts numbers.

[0043] The storage unit 112 includes storage devices such as an HDD (Hard Disk Drive), an SSD (Solid State Drive), and an EEPROM (Electronically Erasable Programmable Read Only Memory). The storage unit 112 stores various computer programs executed by the control unit 111 and various data used by the control unit 111.

[0044] The computer programs stored in the memory unit 112 include a learning processing program PG11 for generating the learning models (observation model MD11 and control model MD12) described later, and an estimation processing program PG12 for estimating the state of the substrate processing device 100 and the control values ​​of actuator A1 using the learning models. These computer programs may be a single computer program or may consist of multiple computer programs. Furthermore, these computer programs may partially utilize existing libraries.

[0045] Computer programs such as the learning processing program PG11 and the estimation processing program PG12, which are stored in the memory unit 112, are provided by a non-temporary recording medium RM10 on which these computer programs are recorded in a readable format. The recording medium RM10 is a portable memory such as a CD-ROM, USB memory, SD (Secure Digital) card, microSD card, or CompactFlash®. The control unit 111 reads various computer programs from the recording medium RM10 using a reading device (not shown in the figure) and stores the read computer programs in the memory unit 112. The computer programs stored in the memory unit 112 may also be provided by communication. In this case, the control unit 111 can acquire the computer programs through the communication unit 115 and store the acquired computer programs in the memory unit 112.

[0046] The memory unit 112 includes a learning model configured to output information about the substrate processing device 100 when time-series data output from the sensor S1 is input. The memory unit 112 stores information that defines the learning model, such as the configuration information of the layers of the learning model, information about the nodes included in each layer, and parameters for weighting and bias between nodes.

[0047] The edge device 110 according to this embodiment includes an observation model MD11 and a control model MD12 as learning models. The observation model MD11 is a model for estimating the state of the substrate processing apparatus 1. The control model MD12 is a model for estimating the control value of actuator A1, which is the control target of the edge device 110. In the example shown in Figure 5, the learning model is configured to include both the observation model MD11 and the control model MD12, but it may also be configured to include only one of them. Also, in the example shown in Figure 5, the configuration includes one observation model MD11 and one control model MD12, but the observation model MD11 may be prepared for each sensor to be observed, and the control model MD12 may be prepared for each actuator to be controlled.

[0048] The input unit 113 is equipped with an interface for connecting a sensor S1. The sensor S1 connected to the input unit 113 is, for example, an RF sensor. The sensor S1 connected to the input unit 113 is not limited to the above; any sensor necessary for observing the process state (performance) can be connected. The time-series data (sensor data) input through the input unit 113 is temporarily stored in the storage unit 112.

[0049] The output unit 114 is equipped with an interface for connecting actuator A1, which is the control target of the edge device 110. The control unit 111 estimates control values ​​using the control model MD12 described above and controls the operation of actuator A1 by outputting control commands based on the estimated control values ​​from the output unit 114 to actuator A1.

[0050] The communication unit 115 is equipped with a communication interface for sending and receiving various types of data with the control device 150. A communication interface compliant with a communication standard such as LAN can be used as the communication interface of the communication unit 115. When data to be transmitted is input from the control unit 111, the communication unit 115 transmits the data to the control device 150, and when it receives data transmitted from the control device 150, it outputs the received data to the control unit 111.

[0051] Figure 6 is a schematic diagram showing an example configuration of the observation model MD11 provided by the edge device 110. The observation model MD11 is configured to output information regarding the state of the substrate processing device 100 on which the sensor S1 is installed (hereinafter referred to as state information) when time-series data (sensor data) obtained from the sensor S1 is input. Any model capable of analyzing time-series data can be adopted as the observation model MD11. For example, the observation model MD11 is a machine learning model including deep learning, and is constructed using RNN (Recurrent Neural Network), LSTM (Long Short-Term Memory), etc. Alternatively, the observation model MD11 may be a learning model based on CNN (Convolutional Neural Network), R-CNN (Region based CNN), YOLO (You Only Look Once), SSD (Single Shot Multibox Detector), GAN (Generative Adversarial Network), SVM (Support Vector Machine), decision tree, etc. Furthermore, the observation model MD11 may be a learning model other than deep learning, such as an autoregressive model, a moving average model, or an autoregressive moving average model.

[0052] The observation model MD11 comprises an input layer MD11a, hidden layers MD11b and MD11c, and an output layer MD11d. In the example in Figure 6, the observation model MD11 is configured to have two hidden layers MD11b and MD11c, but it may also have a configuration with three or more hidden layers.

[0053] The input layer MD11a, the intermediate layers MD11b and MD11c, and the output layer MD11d are provided with one or more nodes. The nodes in each layer are unidirectionally coupled with the nodes in the preceding and succeeding layers with desired weights and biases. The input layer MD11a of the observation model MD11 receives data equal to the number of nodes in the input layer MD11a. In this embodiment, the sensor data input to the nodes of the input layer MD11a is time-series data obtained from the sensor S1. The sensor data input to the input layer MD11a may be a series of time-sequentially continuous measurements, or it may be a graph (image data) plotting the measurements against time.

[0054] The input sensor data is output to the nodes of the first hidden layer MD11b through the nodes of the input layer MD11a. The data input to the first hidden layer MD11b is output to the nodes of the second hidden layer MD11c through the nodes that make up this hidden layer MD11b. During this process, feature quantities of the sensor data are extracted using an activation function that includes weights and biases set between the nodes of each layer.

[0055] The sensor data features extracted by the intermediate layers MD11b and MD11c are output to the output layer MD11d and also taken outside the observation model MD11. The output layer MD11d performs a pre-set calculation using the features input from the second intermediate layer MD11c, and outputs the status information of the substrate processing device 100 as the final calculation result.

[0056] The status information output from the output layer MD11d includes evaluation values ​​that represent the status of the substrate processing apparatus 100. The evaluation values ​​are, for example, information that represents the degree of deterioration of a specific component constituting the substrate processing apparatus 100. Alternatively, the evaluation values ​​may be information that represents whether or not a specific component is faulty. The specific components constituting the substrate processing apparatus 100 include the exhaust system 51, heater power supply 64, gas supply source 73, high-frequency power supply 76, etc.

[0057] The observation model MD11 is trained using any learning algorithm. Supervised learning can be used as the learning algorithm. In this case, a dataset containing sensor data from sensor S1 and the ground truth data that the observation model MD11 should output is used as training data, and the model is trained to output status information of the substrate processing device 100 when sensor data is input. The training data may be provided by the administrator of the substrate processing device 100, etc. For example, sensor data from sensor S1, the date and time when components were replaced, the date and time when a malfunction was discovered, etc., are stored as historical data. Based on this historical data, the sensor data from sensor S1 and the ground truth data indicating whether there was deterioration or a malfunction at the date and time the sensor data was obtained can be provided as training data.

[0058] The observation model MD11 shown as an example in Figure 6 has a configuration comprising an input layer MD11a, intermediate layers MD11b and MD11c, and an output layer MD11d. However, the configuration of the observation model MD11 is not limited to that shown in Figure 6. For example, the observation model MD11 may be a model that does not include intermediate layers MD11b and MD11c, and only defines the input / output relationship (i.e., the relationship between sensor data and state information).

[0059] In this embodiment, supervised learning was described as the learning algorithm for the observation model MD11. However, it is possible to generate the observation model MD11 using any learning algorithm, including unsupervised learning.

[0060] When the control unit 111 of the edge device 110 performs calculations using the observation model MD11, it acquires state information from the output layer MD11d and also acquires feature quantities of sensor data extracted in the intermediate layers MD11b and MD11c. The control unit 111 transmits the acquired state information of the substrate processing device 100 and feature quantities of sensor data to the control device 150, which is a higher-level device of the edge device 110.

[0061] Figure 7 is a schematic diagram showing an example configuration of the control model MD12 provided by the edge device 110. The control model MD12 is configured to output information (hereinafter referred to as control information) regarding the control values ​​of the substrate processing device 100 on which the sensor S1 is installed, when time-series data (sensor data) obtained from the sensor S1 is input. That is, when sensor data is input to the input layer MD12a of the control model MD12, calculations to extract feature quantities of the sensor data are performed in the intermediate layers MD12b and MD12c. The feature quantities obtained from the intermediate layers MD12b and MD12c are output to the output layer MD12d and also taken out of the control model MD12. The output layer MD12d performs a predetermined calculation using the feature quantities input from the second intermediate layer MD12c, and outputs the control information of the substrate processing device 100 as the final calculation result. The feature quantities of the sensor data output from the intermediate layer MD12c and the control information output from the output layer MD12d are input to the control unit 111. The control information output from the output layer MD12d includes a control value for controlling at least one component of the substrate processing device 100.

[0062] The control model MD12, like the observation model MD11, may be trained using supervised learning or unsupervised learning. Furthermore, the control model MD12 may be trained using reinforcement learning. For example, the value in reinforcement learning can be learned to maximize the total rewards obtained over time by assigning rewards according to the state of the substrate processing device 100. For instance, Q-learning, a type of reinforcement learning, learns the value Q for selecting an action (control value) under certain environmental conditions. At the start of Q-learning, the correct value Q for a given combination of the state of the substrate processing device 100 and an action (control value) is unknown. Therefore, the correct value Q is learned by selecting various control values ​​under certain work data, calculating the total reward based on the reward given for the action (control based on the control value), and selecting better control values.

[0063] The control model MD12 is not limited to the model described above; it may be any other model capable of analyzing time-series data. For example, the control model MD12 may be any learning model other than deep learning, such as an autoregressive model, a moving average model, or an autoregressive moving average model. Furthermore, the configuration of the control model MD12 is not limited to that shown in Figure 7. For example, the control model MD12 may be a model that does not include hidden layers MD12b and MD12c, and only defines the input-output relationship (i.e., the relationship between sensor data and control information).

[0064] When the control unit 111 of the edge device 110 performs calculations using the control model MD12, it acquires control information from the output layer MD12d and also acquires feature quantities of sensor data calculated in the intermediate layers MD12b and MD12c. The control unit 111 transmits the acquired control information of the substrate processing device 100 and feature quantities of sensor data to the control device 150, which is a higher-level device of the edge device 110. The control unit 111 also controls the operation of actuator A1 based on the control information acquired from the control model MD12.

[0065] In this embodiment, features are extracted from the intermediate layers MD11c and MD12c, but the final calculation results obtained from the output layers MD11d and MD12d may be considered as the features of the sensor data. Alternatively, the control unit 111 may directly extract features from the sensor data. Appropriate statistical processing such as peak detection and interval averaging may be used for feature extraction. The control unit 111 may also extract features by detecting anomalies appearing in the sensor data and weighting the data of the detected anomalies. Alternatively, the control unit 111 may extract snapshots of time-series data containing anomalies appearing in the sensor data as features.

[0066] Figures 5 to 7 illustrate the internal configuration of edge device 110, and the internal configuration of edge devices 120 and 130 is similar. Specifically, edge devices 120 and 130 each have an observation model and a control model, and estimate the state of the substrate processing apparatus 100 and the control values ​​of actuators A1 and A2 based on sensor data input from sensors S2 and S3. Edge devices 120 and 130 transmit the state information of the substrate processing apparatus 100 and the control information of actuators A1 and A2 obtained from the observation model and control model, as well as the characteristic quantities of the sensor data output from sensors S2 and S3, to the control device 150, which is a higher-level device of edge devices 120 and 130. Furthermore, edge devices 120 and 130 control the operation of actuators A2 and A3, respectively, based on the control information acquired from the control model.

[0067] Figure 8 is a block diagram showing the internal configuration of the control device 150 provided by the substrate processing apparatus 100. The control device 150 is a dedicated or general-purpose computer installed inside the substrate processing apparatus 100, and includes a control unit 151, a storage unit 152, a first communication unit 153, a second communication unit 154, an operation unit 155, a display unit 156, and the like. The control device 150 collects data (feature quantities of sensor data) transmitted from edge devices 110 to 130 and stores it in the database DB20 in the storage unit 152. Based on the medium-term data stored in the database DB20, the control device 150 generates a second learning model that absorbs individual differences in the sensors.

[0068] The control unit 151 includes a CPU, ROM, RAM, etc. The ROM in the control unit 151 stores control programs and the like that control the operation of each hardware component of the control device 150. The CPU in the control unit 151 reads and executes the control programs stored in the ROM and various computer programs stored in the memory unit 152 to control the operation of each hardware component.

[0069] The control unit 151 is not limited to the configuration described above, and may be one or more control circuits or arithmetic circuits equipped with a GPU, FPGA, DSP, quantum processor, volatile or non-volatile memory, etc. Furthermore, the control unit 151 may include functions such as a clock for outputting date and time information, a timer for measuring the elapsed time from the start of measurement to the end of measurement, and a counter for counting numbers.

[0070] The storage unit 152 includes a storage device such as an HDD, SSD, or EEPROM. The storage unit 152 also includes the database DB20 described above. Figure 9 is a conceptual diagram showing an example configuration of the database DB20. The database DB20 stores date and time information and sensor data feature quantities associated with the identification information (device ID) of the edge devices 110 to 130. Furthermore, the database DB20 may also store state information and control information of the substrate processing device 100.

[0071] The storage unit 152 stores the database DB20, various computer programs executed by the control unit 151, and various data used by the control unit 151.

[0072] The computer program stored in the memory unit 152 includes a learning processing program PG21 for generating a second learning model, and an estimation processing program PG22 for estimating the state and control values ​​of the substrate processing device 100 using the second learning model. The computer program stored in the memory unit 152 is provided on a non-temporary recording medium RM20 on which the computer program is recorded in a readable format. Alternatively, the computer program stored in the memory unit 152 may be provided via communication.

[0073] The memory unit 152 includes a second learning model configured to output information about the substrate processing device 100 when sensor data features are input. The memory unit 152 stores information defining the second learning model, such as the configuration information of the layers of the second learning model, information about the nodes included in each layer, and information about the weights and biases between nodes.

[0074] The control device 150 includes an observation model MD21 and a control model MD22 as second learning models. The observation model MD21 is a model for estimating the state of the substrate processing apparatus 100. The control model MD22 is a model for estimating the control values ​​used in the substrate processing apparatus 100. The configurations of the observation model MD21 and the control model MD22 are the same as those of the observation model MD11 and the control model MD12 provided by the edge devices 110 to 130, so a detailed explanation is omitted.

[0075] In this embodiment, the second learning model is configured to include both the observation model MD21 and the control model MD22, but it may also be configured to include only one of them. Furthermore, in this embodiment, the second learning model is configured to include one observation model MD21 and one control model MD22, but the observation model MD21 may be prepared for each observation target, and the control model MD22 may be prepared for each controlled target.

[0076] The first communication unit 153 is equipped with a communication interface for sending and receiving various types of data with the edge devices 110 to 130. A communication interface compliant with a communication standard such as LAN can be used as the communication interface of the first communication unit 153. When the first communication unit 153 receives input of data to be sent from the control unit 151, it sends the data to the destination edge devices 110 to 130, and when it receives data sent from the edge devices 110 to 130, it outputs the received data to the control unit 151.

[0077] The second communication unit 154 is equipped with a communication interface for sending and receiving various types of data. The communication interface provided by the second communication unit 154 is, for example, a communication interface conforming to the LAN communication standards used in WiFi (registered trademark) and Ethernet (registered trademark). When the second communication unit 154 receives data to be transmitted from the control unit 151, it transmits the data to be sent to the specified destination. Also, when the second communication unit 154 receives data transmitted from an external device, it outputs the received data to the control unit 151.

[0078] The operation unit 155 is equipped with operating devices such as a touch panel, keyboard, and switches, and accepts various operations and settings from administrators, etc. The control unit 151 performs appropriate control based on the various operation information provided by the operation unit 155, and stores setting information in the storage unit 152 as needed.

[0079] The display unit 156 is equipped with a display device such as a liquid crystal monitor or an organic EL (Electro-Luminescence) display and displays information that should be notified to administrators, etc., in response to instructions from the control unit 151.

[0080] Next, we will describe the configuration of the device group server 200. Figure 10 is a block diagram showing the internal configuration of the device group server 200. The device group server 200 includes a control unit 201, a storage unit 202, a communication unit 203, an operation unit 204, a display unit 205, and the like. The device group server 200 collects data transmitted from multiple substrate processing devices 100 and stores it in a database DB 30 located in the storage unit 202. Based on the long-term data stored in the database DB 30, the device group server 200 generates a third learning model that absorbs individual differences between devices.

[0081] The control unit 201 includes a CPU, ROM, RAM, etc. The ROM in the control unit 201 stores control programs that control the operation of each hardware component of the device group server 200. The CPU in the control unit 201 reads and executes the control programs stored in the ROM and various computer programs stored in the memory unit 202 to control the operation of each hardware component.

[0082] The control unit 201 is not limited to the configuration described above, and may be one or more control circuits or arithmetic circuits equipped with a GPU, FPGA, DSP, quantum processor, volatile or non-volatile memory, etc. Furthermore, the control unit 201 may include functions such as a clock for outputting date and time information, a timer for measuring the elapsed time from the start of measurement to the end of measurement, and a counter for counting numbers.

[0083] The storage unit 202 includes a storage device such as an HDD, SSD, or EEPROM. The storage unit 202 also includes the database DB30 described above. Figure 11 is a conceptual diagram showing an example configuration of the database DB30. The database DB30 stores date and time information and characteristic quantities of sensor data, associated with the identifier (device ID) of the substrate processing device 100. The database DB30 may also store state information indicating the state of the substrate processing device 100, and control information used to control the substrate processing device 100.

[0084] The storage unit 202 stores the database DB30, various computer programs executed by the control unit 201, and various data used by the control unit 201.

[0085] The computer program stored in the memory unit 202 includes a learning processing program PG31 for generating a third learning model, and an estimation processing program PG32 for estimating the state and control values ​​of the substrate processing device 100. The computer program stored in the memory unit 202 is provided by a non-temporary recording medium RM30 on which the computer program is recorded in a readable format. Alternatively, the computer program stored in the memory unit 202 may be provided by communication.

[0086] The memory unit 202 includes a third learning model configured to output information about the substrate processing device 100 when sensor data features are input. The memory unit 202 stores information defining the third learning model, such as the configuration information of the layers of the third learning model, information about the nodes included in each layer, and information about the weights and biases between nodes.

[0087] The device group server 200 includes an observation model MD31 and a control model MD32 as third learning models. The observation model MD31 is a model for estimating the state of the substrate processing device 100. The control model MD32 is a model for estimating the control values ​​used in the substrate processing device 100. The configurations of the observation model MD31 and the control model MD32 are the same as those of the observation model MD21 and control model MD22 provided in the control device 150, so a detailed explanation is omitted.

[0088] In this embodiment, the third learning model is configured to include both the observation model MD31 and the control model MD32, but it may also be configured to include only one of them. Furthermore, in this embodiment, the third learning model is configured to include one observation model MD31 and one control model MD32, but the observation model MD31 may be prepared for each observation target, and the control model MD32 may be prepared for each controlled target.

[0089] The communication unit 203 is equipped with a communication interface for sending and receiving various types of data. The communication interface provided by the communication unit 203 is, for example, a communication interface conforming to LAN communication standards used in WiFi (registered trademark) and Ethernet (registered trademark). When the communication unit 203 receives data to be transmitted from the control unit 201, it transmits the data to the specified destination. Also, when the communication unit 203 receives data transmitted from an external device, it outputs the received data to the control unit 201.

[0090] The operation unit 204 is equipped with operating devices such as a touch panel, keyboard, and switches, and accepts various operations and settings from administrators, etc. The control unit 201 performs appropriate control based on the various operation information provided by the operation unit 204, and stores setting information in the storage unit 202 as needed.

[0091] The display unit 205 is equipped with a display device such as an LCD monitor or an organic EL display and displays information that should be notified to administrators, etc., in response to instructions from the control unit 201.

[0092] In the example shown in Figure 10, the device group server 200 is configured to include an operation unit 204 and a display unit 205. However, the operation unit 204 and the display unit 205 are not essential components of the device group server 200. If the operation unit 204 is not provided, the device group server 200 can receive operations from an external computer that is connected via a communication unit 203. Similarly, if the display unit 205 is not provided, the device group server 200 can transmit information to be communicated to an administrator or other relevant party from the communication unit 203 to an external computer and have the information displayed on the external computer.

[0093] The operation of the substrate processing system will be described below. In the substrate processing system according to this embodiment, during the learning phase before operation begins, edge devices 110 to 130 generate a first learning model (observation model MD11 and control model MD12).

[0094] Figure 12 is a flowchart showing the procedure for generating the first learning model by the edge device 110. The control unit 111 of the edge device 110 collects sensor data output from the sensor S1 in a time series via the input unit 113 (step S101). The sensor data collection period is, for example, one month. When acquiring sensor data, the control unit 111 receives state information of the substrate processing device 100 from an external source and obtains control values ​​to be output to the substrate processing device 100 from the output unit 114. The control values ​​of the substrate processing device 100 used when the learning of the control model MD12 is not yet complete are determined, for example, by referring to a pre-set recipe. These state information and control values, along with the sensor data, are stored in the storage unit 112 as training data when learning the observation model MD11 and the control model MD12.

[0095] When the training data necessary for learning is obtained, the control unit 111 selects a set of training data from the training data stored in the memory unit 112 (step S102). The control unit 111 inputs the sensor data included in the selected set of training data into the observation model MD11 and the control model MD12, respectively, and performs calculations for the observation model MD11 and the control model MD12 (step S103). Before learning begins, it is assumed that initial values ​​have been set for the model parameters of the observation model MD11 and the control model MD12.

[0096] The control unit 111 evaluates the calculation results of the observation model MD11 and the control model MD12 (step S104) and determines whether the learning of the observation model MD11 and the control model MD12 is complete (step S105). The control unit 111 can evaluate the calculation results using an error function (also called an objective function, loss function, or cost function) based on the calculation results by the models and the state or control values ​​included as ground truth data. For example, in the process of optimizing (minimizing or maximizing) the error function using a gradient descent method such as the steepest descent method, the control unit 111 determines that the learning of the observation model MD11 and the control model MD12 is complete if the error function becomes below (or above) a threshold.

[0097] If learning is not complete (S105:NO), that is, if learning of either the observation model MD11 or the control model MD12 is not complete, the control unit 111 updates the parameters of the model that has not been learned (such as weights and biases between nodes) (step S106) and returns the process to step S102. The control unit 111 can update the parameters in the model using a backpropagation method, which sequentially updates the weights and biases between nodes from the output layers MD11d and MD12d toward the input layers MD11a and MD12a.

[0098] If it is determined that learning is complete (S105:YES), the learned observation model MD11 and control model MD12 are obtained, and the control unit 111 stores them in the memory unit 112 as the first learned model (step S107).

[0099] Figure 12 illustrates the procedure for generating the first learning model using edge device 110, but the same generation procedure can be used to generate the first learning model applicable to each of the edge devices 120 and 130.

[0100] In this embodiment, the substrate processing system transitions to the operation phase after a first learning model is generated in each edge device 110 to 130. In the operation phase, the substrate processing system performs the following processes.

[0101] Figure 13 is a flowchart showing the procedure of processing performed inside the substrate processing device 100 during the operation phase. When the control unit 111 of the edge device 110 (120, 130) provided in the substrate processing device 100 acquires sensor data output in time series from the sensor S1 via the input unit 113 (step S121), it executes the model by inputting the acquired sensor data into the observation model MD11 or the control model MD12 (step S122).

[0102] During the execution of the model, the control unit 111 extracts feature quantities of sensor data from the intermediate layers MD11c and MD12c of each model MD11 and MD12 (step S123). The control unit 111 can, for example, extract feature quantities from the intermediate layers MD11c and MD12c. Alternatively, the control unit 111 may consider the final calculation results obtained from the output layers MD11d and MD12d as feature quantities of sensor data, or it may directly extract feature quantities from the sensor data.

[0103] The control unit 111 transmits the extracted feature quantities to the control device 150 along with the estimated state information and control information obtained from each model MD11 and MD12 (step S124). The control unit 111 also controls the actuator A1 based on the state information and control information estimated by each model MD11 and MD12 (step S125). The control unit 111 only needs to execute the processes in steps S122 to S125 each time it acquires sensor data in step S122.

[0104] The control unit 151 of the control device 150 receives the feature quantities transmitted from the edge device 110 (120, 130) from the first communication unit 153 (step S126) and stores them in the database DB20 (step S127).

[0105] The control unit 151 determines whether the feature collection period has ended (step S128). The collection period is, for example, 6 months from the start of feature collection. Alternatively, the determination of whether the collection period has ended may be made based on the number of features accumulated in the database DB20. If the collection period has not ended (S128: NO), the control unit 151 returns to step S126 and repeats the process of receiving features and accumulating them in the database DB20.

[0106] If the data collection period ends (S128: YES), the observation model MD21 and the control model MD22 are trained and the models are created (step S129). For example, the control unit 151 can create the observation model MD21 by training it using a set of feature quantities and state information stored in the database DB20 as training data. The control unit 151 can also create the control model MD22 by training it using a set of feature quantities and control values ​​stored in the database DB20 as training data. The procedure for creating the models is the same as the procedure for creating the observation model MD11 and the control model MD12.

[0107] After creating the model, if the control unit 151 receives new features from the edge device 110 (120, 130), it inputs the received features into the observation model MD21 or the control model MD22 and runs the model (step S130).

[0108] Based on the results of step S130, the control unit 151 determines whether or not the model used in the edge device 110 (120, 130) needs to be updated (step S131). Since the control device 150 creates a model based on accumulated medium-term data (for example, data in 6-month increments), the control unit 151 can determine the deviation from the trend shown by the model by running the model based on newly acquired features. If the deviation from the trend shown by the model is greater than or equal to a threshold, the control unit 151 determines that there is an abnormality in the model provided by the edge device 110 (120, 130). If the control unit 151 determines that there is no abnormality in the model, it determines that a model update is not necessary (S131: NO), and the control unit 151 returns to step S130.

[0109] If the control unit 151 determines that there is an abnormality in the model, it determines that the model needs to be updated (S131:YES) and sends a model retraining instruction to the edge device 110 (120,130) (step S132).

[0110] The control unit 111 of the edge device 110 (120, 130) determines whether or not it has received a relearning instruction transmitted from the control device 150 (step S133). If it determines that it has not received a relearning instruction (S133: NO), the control unit 111 returns to step S121 and repeatedly executes the processes in steps S121 to S125.

[0111] If a relearning instruction is received (S133: YES), the control unit 111 performs relearning (step S134). The control unit 111 can retrain the observation model MD11 by performing additional learning using, for example, sensor data obtained from sensor S1 and state information of the substrate processing device 100 as training data. The control unit 111 can also retrain the control model MD12 by performing additional learning using, for example, sensor data obtained from sensor S1 and control values ​​used in the substrate processing device 100 as training data. Instead of retraining the observation model MD11 and control model MD12 by additional learning, the observation model MD11 and control model MD12 may be retrained from the beginning using the above training data.

[0112] In this flowchart, the control device 150 is configured to send a retraining instruction to the edge devices 110 (120, 130) when it determines that a model update is necessary. However, correction values ​​to correct the calculation results of each model MD11 and MD12 may also be sent to the edge devices 110 (120, 130). For example, correction values ​​can be calculated from the error between the prediction results from the second learning model (observation model MD21 and control model MD22) and the actual measurement results.

[0113] Figure 14 is a flowchart showing the procedure for processing performed between the substrate processing device 100 and the device group server 200 during the operation phase. As described above, the control device 150 of the substrate processing device 100 creates the observation model MD21 and the control model MD22 based on the feature quantities obtained from the edge devices 110 to 130 during the operation phase. After the model is created, if the control unit 151 of the control device 150 receives new feature quantities from the edge devices 110 (120, 130) (step S141), it inputs the received feature quantities into the observation model MD21 or the control model MD22 and runs the model (step S142).

[0114] The control unit 111 extracts sensor data features from the intermediate layers of each model MD21 and MD22 during the execution of the model (step S143), and transmits the extracted features, along with the estimated state information and control information obtained from each model MD21 and MD22, to the device group server 200 (step S144). In this embodiment, the sensor data features are extracted from the intermediate layers of each model MD21 and MD22 and transmitted to the device group server 200. However, the final calculation results obtained from the output layers of each model MD21 and MD22 may be considered as sensor data features and transmitted to the device group server 200.

[0115] The control unit 201 of the device group server 200 receives the feature quantities transmitted from the substrate processing device 100 from the communication unit 203 (step S145) and stores them in the database DB 30 (step S146).

[0116] The control unit 201 determines whether the feature collection period has ended (step S147). The collection period is, for example, 2 to 3 years from the start of feature collection. Alternatively, the determination of whether the collection period has ended may be made based on the number of features accumulated in the database DB30. If the collection period has not ended (S147: NO), the control unit 201 returns to step S145 and repeats the process of receiving features and accumulating them in the database DB30.

[0117] If the data collection period ends (S147:YES), the observation model MD31 and the control model MD32 are trained and the models are created (step S148). For example, the control unit 201 can create the observation model MD31 by training it using a set of feature quantities and state information stored in the database DB30 as training data. The control unit 201 can also create the control model MD32 by training it using a set of feature quantities and control values ​​stored in the database DB30 as training data. The procedure for creating the models is the same as the procedure for creating the observation model MD11 and the control model MD12.

[0118] After creating the model, if the control unit 201 receives new feature quantities from the substrate processing device 100, it inputs the received feature quantities into the observation model MD31 or the control model MD32 and runs the model (step S149).

[0119] Based on the results of step S149, the control unit 201 determines whether or not it is necessary to update the model used in the edge devices 110 (120, 130) of the substrate processing device 100 (step S150). Since the device group server 200 creates a model based on accumulated long-term data (for example, data over a period of 2 to 3 years), the control unit 201 can determine the deviation from the trend shown by the model by running the model based on newly acquired features. If the deviation from the trend shown by the model is greater than or equal to a threshold, the control unit 201 determines that there is an abnormality in the model provided by the edge devices 110 (120, 130). If the control unit 201 determines that there is no abnormality in the model, it determines that an update to the model is not necessary (S150: NO), and the control unit 201 returns to step S149.

[0120] If the control unit 201 determines that there is an abnormality in the model, it determines that the model needs to be updated (S150: YES) and sends a model retraining instruction to the substrate processing device 100 (step S151).

[0121] The control unit 151 of the control device 150 of the substrate processing apparatus 100 determines whether or not it has received a relearning instruction transmitted from the apparatus group server 200 (step S152). If it determines that it has not received a relearning instruction (S152: NO), the control unit 151 returns to step S141 and repeatedly executes the processes of steps S141 to S144.

[0122] If a relearning instruction is received (S152: YES), the control unit 151 instructs the edge device 110 (120, 130) to perform relearning (step S153). The control unit 111 of the edge device 110 (120, 130) can retrain the observation model MD11 by performing additional learning using, for example, sensor data obtained from sensor S1 and state information of the substrate processing device 100 as training data. The control unit 111 can also retrain the control model MD12 by performing additional learning using, for example, sensor data obtained from sensor S1 and control values ​​used in the substrate processing device 100 as training data. Instead of retraining the observation model MD11 and control model MD12 by additional learning, the observation model MD11 and control model MD12 may be retrained from the beginning using the above training data.

[0123] In this flowchart, when a retraining instruction is given from the device group server 200, the edge devices 110 (120, 130) are configured to retrain the observation model MD11 and the control model MD12. However, the control device 150 may be configured to retrain the observation model MD21 and the control model MD22. The control unit 151 of the control device 150 can retrain the observation model MD21 using the feature quantities and state information stored in the database DB20 as training data. The control unit 151 can also retrain the control model MD22 using the feature quantities and control values ​​stored in the database DB20 as training data. The control unit 151 may retrain the observation model MD21 and the control model MD22 through additional training, or it may retrain the observation model MD21 and the control model MD22 from the beginning using the above training data.

[0124] In this flowchart, the device group server 200 is configured to send a retraining instruction to the substrate processing device 100 when it determines that a model update is necessary. However, correction values ​​to correct the model's calculation results may also be sent to the substrate processing device 100. For example, correction values ​​can be calculated from the error between the prediction results from the third learning model (observation model MD31 and control model MD32) and the actual measurement results.

[0125] As described above, in this embodiment, each edge device 110 to 130 can create a high-resolution model (observation model MD11 and control model MD12) based on sensor data. Furthermore, the edge devices 110 to 130 may create a model that estimates the degradation of components provided by the substrate processing device 100 based on the acquired sensor data.

[0126] Each edge device 110-130 transmits the features extracted by each model to the control device 150 without transmitting sensor data, thereby reducing the network load between the edge devices 110-130 and the control device 150. The control device 150 can create a medium-term trend model (observation model MD21 and control model MD22) based on the features of the sensor data.

[0127] Each substrate processing unit 100 transmits the features extracted by each model to the device group server 200, thereby reducing the network load between the substrate processing units 100 and the device group server 200. Based on the features transmitted from the substrate processing units 100, the device group server 200 can create long-term trend models (observation model MD31 and control model MD32) that absorb individual differences between devices.

[0128] Furthermore, if a new substrate processing device (not shown) is introduced into the substrate processing system, the trained first learning model (observation model MD11 and control model MD12) may be deployed to each edge device 110-130. In addition, the trained second learning model (observation model MD21 and control model MD22) may be deployed to the control device 150 of each substrate processing device 100.

[0129] (Embodiment 2) Embodiment 2 describes a configuration that evaluates the completeness and health of the first learning model and outputs the evaluation results. The system configuration and the internal configuration of the substrate processing device 100 and the device group server 200 are the same as in Embodiment 1, so their explanation will be omitted.

[0130] The substrate processing device 100 evaluates the completeness and health of the first learning models (observation model MD11 and control model MD12) provided by the edge devices 110-130 at appropriate timings during the learning phase or operation phase, and outputs the evaluation results.

[0131] The substrate processing device 100 is equipped with evaluation datasets for evaluating the first learning models (observation model MD11 and control model MD12). For example, to evaluate the observation model MD11 of the edge device 110, a set including sensor data from sensor S1 and the ground truth data that the observation model MD11 should output when sensor data from sensor S1 is input can be used as the evaluation dataset. Similarly, to evaluate the control model MD12 of the edge device 110, a set including sensor data from sensor S1 and the ground truth data that the control model MD12 should output when sensor data from sensor S1 is input can be used as the evaluation dataset. The same applies to evaluation datasets for evaluating edge devices 120 and 130.

[0132] The substrate processing device 100 can evaluate the completeness and health of the first learning model based on the difference between the estimated values ​​obtained when sensor data included in the evaluation dataset is input to the observation model MD11 and the control model MD12, and the ground truth data included in the dataset.

[0133] When the substrate processing device 100 evaluates the completeness and health of the first learning model, it displays the evaluation results on the display unit 156. Figure 15 is a schematic diagram showing an example of the display of evaluation results. In the example in Figure 15, the results of evaluating the completeness and health of the observation model MD11 and control model MD12 equipped on each of the edge devices 110 to 130 are shown. In each graph, the indices A, B, and C represent edge devices 110, 120, and 130, respectively. The upper graph shows how the completeness of the observation model MD11 and control model MD12 increases with the number of learning iterations. The lower graph shows the health of the observation model MD11 and control model MD12 equipped on each edge device 110, 120, and 130 at the time of evaluation.

[0134] Thus, in Embodiment 2, the performance of each learning model can be displayed in a list, so if the completeness or health of the observation model MD11 and the control model MD12 is insufficient, the administrator can improve the completeness and health of the observation model MD11 and the control model MD12 by issuing a retraining instruction through the operation unit 155.

[0135] The embodiments disclosed herein should be considered in all respects to be illustrative and not restrictive. The scope of the invention is indicated by the claims, not in the sense described above, and all modifications within the sense and scope equivalent to the claims are intended. [Explanation of Symbols]

[0136] 110, 120, 130 Edge Devices 111 Control Unit 112 Storage section 113 Input section Output section of 114 115 Communications Department 150 control devices 151 Control Unit 152 Storage section 153 First Communications Department 154 Second Communications Department 155 Operation section 156 Display section 200-device server group 201 Control Unit 202 Storage section 203 Communications Department 204 Operation section 205 Display section MD11, MD21, MD31 observation models MD12, MD22, MD32 control models

Claims

1. A process for acquiring time-series data from multiple types of sensors with different sampling periods, which are installed in a substrate processing apparatus, The process involves individually training a first learning model that outputs information related to the substrate processing device when time-series data from each sensor is input, using time-series data with different sampling periods for each sensor. The process involves inputting time-series data output from each sensor into a corresponding first trained model, and outputting estimation results based on the information obtained from the first trained model. Information processing methods including

2. The first learning model for each sensor is deployed for each edge device corresponding to that sensor. This includes the step of performing the learning process of the first learning model and the estimation process by the first learning model on each edge device. The information processing method according to claim 1.

3. The first learning model described above is An observation model for estimating the state of the substrate processing apparatus based on time-series data from the aforementioned sensor, and A control model for estimating the control values ​​of the substrate processing apparatus based on time-series data from the aforementioned sensor. The information processing method according to claim 1, comprising at least one of the following.

4. A step of extracting a first feature from the aforementioned time series data, A step of outputting the extracted first feature quantity to the first device in the substrate processing apparatus. The information processing method according to claim 1, further comprising:

5. A process for storing the first feature quantity extracted from the time-series data, Based on the stored first feature quantity, a process is performed to train a second learning model that outputs information related to the substrate processing device when the first feature quantity is input. The process involves inputting the newly acquired first features into the second trained model and outputting estimation results based on the information obtained from the second trained model. The information processing method according to claim 4, further comprising the step of performing the first device.

6. A process in which the first device outputs a retraining instruction for the first learning model based on the estimation results using the second learning model. The information processing method according to claim 5, further comprising:

7. A step of outputting a correction value to correct the calculation result of the first learning model based on the calculation result of the second learning model. The information processing method according to claim 5, further comprising:

8. A step of extracting a second feature quantity of the time-series data for each substrate processing apparatus, A step of outputting the extracted second feature quantity to a second device outside the substrate processing device. The information processing method according to claim 5, further comprising:

9. A process for storing the second feature quantity extracted for each substrate processing apparatus, Based on the stored second feature quantity, a process is performed to train a third learning model that outputs information related to the substrate processing device when the second feature quantity is input. The process involves inputting the newly acquired second feature into the third learning model after training, and outputting an estimation result based on the information obtained from the third learning model. The information processing method according to claim 8, further comprising the step of performing the second device.

10. A process in which the second device outputs a retraining instruction for the first or second learning model based on the estimation results using the third learning model. The information processing method according to claim 9, further comprising:

11. A step of outputting a correction value to correct the calculation result of the first learning model or the second learning model based on the calculation result of the third learning model. The information processing method according to claim 9, further comprising:

12. When installing a new substrate processing apparatus, the process involves introducing the previously learned first learning model into the new substrate processing apparatus. The information processing method according to claim 1, further comprising:

13. The process of displaying the performance of each learning model in a list. The information processing method according to claim 1, further comprising:

14. A substrate processing apparatus is provided with an acquisition unit that acquires time-series data from each of several types of sensors with different sampling periods, A learning unit individually trains a first learning model that outputs information related to the substrate processing device when time-series data from each sensor is input, using time-series data with different sampling periods for each sensor. An estimation unit inputs time-series data output from each sensor into a corresponding first learned model and outputs estimation results based on the information obtained from the first learned model. An information processing device equipped with the following features.

15. Multiple substrate processing devices, comprising an edge device connected to a sensor and a higher-level device connected to the edge device, which perform substrate processing inside a chamber, A group of device servers that are communicably connected to the aforementioned plurality of substrate processing devices. Includes, The edge device is An acquisition unit that acquires time-series data from the aforementioned sensor, A first learning unit performs training on a first learning model that outputs information about a substrate processing device on which the sensor is installed, based on the acquired time-series data, when time-series data from the sensor is input. A first estimation unit inputs time-series data from the sensor to the first learned model after training and outputs estimation results based on the information obtained from the first learned model. An output unit that outputs a first feature quantity extracted from the aforementioned time-series data to the aforementioned higher-level device. Equipped with, The aforementioned higher-level device is A first feature quantity storage unit that stores a first feature quantity input from the edge device, A second learning unit performs training on a second learning model that outputs information related to the substrate processing apparatus when the first feature quantity is input, based on the first feature quantity that has been stored. A second estimation unit inputs newly acquired first features into the second learning model after training and outputs estimation results based on the information obtained from the second learning model, A transmission unit that transmits the second characteristic quantity of the time-series data extracted for each substrate processing device to the device group server. Equipped with, The aforementioned group of devices server, A second feature quantity storage unit that stores the second feature quantity received from the aforementioned higher-level device, A third learning unit performs training on a third learning model that outputs information related to the substrate processing apparatus when a second feature quantity is input, based on the second feature quantity that has been stored. A third estimation unit inputs newly acquired second features into the third learning model after training and outputs estimation results based on the information obtained from the third learning model. Equipped with PCB processing system.

16. The aforementioned higher-level device and the aforementioned device group server are A determination unit that determines whether or not the first learning model needs to be updated based on the estimation results obtained by the learning models each possesses, If it is determined that an update is necessary, the instruction unit instructs the edge device to retrain the first learning model. The substrate processing system according to claim 15, comprising:

Images