Production Equipment Monitoring System and Production Equipment Monitoring Method

Abstract

To readily detect a facility abnormality attributable to plural factors.SOLUTION: A production facility monitoring system 20 to be disclosed includes: an abnormality index determination unit 23 that determines an abnormality index (a) for a production facility 10 on the basis of a feature quantity obtained from facility information on the production facility 10; a relation degree determination unit 24 that determines a relation degree D between each of plural abnormal conditions which may occur in the production facility 10 and an observed condition of the production facility 10; a detection threshold determination unit 25 that determines a single detection threshold (th) which is used to detect an abnormality degree A of the production facility 10, on the basis of plural abnormality thresholds at which are thresholds of the abnormality index (a) associated with plural abnormal conditions, and the relation degree D; and an abnormality degree detection unit 26 that detects the abnormality degree A of the production facility 10 on the basis of the abnormality index (a) and the detection threshold (th).SELECTED DRAWING: Figure 2

Description

Technical Field

[0001] The present disclosure relates to a production facility monitoring system and a production facility monitoring method.

Background Art

[0002] Conventionally, a method for determining an abnormality of a production facility based on parameters of the production facility has been known (for example, Patent Document 1). In the method of Patent Document 1, a plurality of data are sampled for parameters of a production facility in a normal state (specifically, a semiconductor manufacturing apparatus), a Mahalanobis space is created from the sampled data group, a Mahalanobis distance is calculated from a group of measured values of parameters of the production facility in an operating state, and an abnormality determination of the production facility is performed based on the Mahalanobis distance.

Prior Art Documents

Patent Documents

[0003]

Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] By the way, a plurality of factors may be involved in the occurrence of an abnormality in a production facility. In such a case, it is difficult to easily detect an abnormality of a production facility by a known method including the method of Patent Document 1. In such a situation, one of the objects of the present disclosure is to easily detect an equipment abnormality caused by a plurality of factors.

Means for Solving the Problems

[0005] One aspect of the present disclosure relates to a production facility monitoring system. The production facility monitoring system includes an abnormality index determination unit that determines an abnormality index of the production facility based on a feature amount obtained from facility information of the production facility, a relevance determination unit that determines a relevance between each of a plurality of abnormal states that may occur in the production facility and an observation state of the production facility, a detection threshold determination unit that determines a single detection threshold for detecting an abnormality degree of the production facility based on a plurality of abnormality thresholds that are thresholds of the abnormality index corresponding to each of the plurality of abnormal states and the relevance, and an abnormality degree detection unit that detects the abnormality degree of the production facility based on the abnormality index and the detection threshold.

[0006] Another aspect of the present disclosure relates to a production facility monitoring method. The production facility monitoring method includes an abnormality index determination step of determining an abnormality index of the production facility based on a feature amount obtained from facility information of the production facility, a relevance determination step of determining a relevance between each of a plurality of abnormal states that may occur in the production facility and an observation state of the production facility, a detection threshold determination step of determining a single detection threshold for detecting an abnormality degree of the production facility based on a plurality of abnormality thresholds that are thresholds of the abnormality index corresponding to each of the plurality of abnormal states and the relevance, and an abnormality degree detection step of detecting the abnormality degree of the production facility based on the abnormality index and the detection threshold.

Advantages of the Invention

[0007] According to the present disclosure, it is possible to easily detect equipment abnormalities caused by a plurality of factors.

Brief Description of the Drawings

[0008]

Figure 1

Figure 2

Figure 3

Embodiments for Carrying Out the Invention

[0009] Embodiments of a production facility monitoring system and a production facility monitoring method according to the present disclosure will be described below with examples. However, the present disclosure is not limited to the examples described below. In the following description, specific numerical values and materials may be exemplified, but other numerical values and materials may be applied as long as the effects of the present disclosure can be obtained.

[0010] (Production Facility Monitoring System) The production facility monitoring system according to the present disclosure includes an abnormality index determination unit, a relevance determination unit, a detection threshold determination unit, and an abnormality degree detection unit.

[0011] The abnormality index determination unit determines an abnormality index of the production facility based on a feature amount obtained from the facility information of the production facility. The production facility broadly refers to a facility that produces something. Specific examples of the production facility can include plasma processing devices such as a plasma cleaner and a plasma dicer. The facility information may be, for example, information on a control signal used for controlling the production facility (hereinafter also referred to as control information), or information on a signal output by a sensor provided in the production facility corresponding to the control information (hereinafter also referred to as sensor information). The feature amount may be obtained by performing extraction or calculation according to a predetermined rule based on such facility information. The abnormality index is an index that can be used to determine how much the production facility deviates from the normal state, and it is also possible to immediately detect an abnormality of the production facility from this. However, in the present disclosure, an abnormality of the production facility is not detected only using the abnormality index.

[0012] The relevance determination unit determines the relevance between each of a plurality of abnormal states that may occur in the production facility and the observed state of the production facility. The abnormal state of the production facility refers to a state in which equipment abnormality has occurred due to a certain factor. The observed state of the production facility refers to a state observed in the operating production facility. The relevance refers to the degree indicating how closely an abnormal state and the observed state are related in the production facility. For example, if the relevance between a certain abnormal state (here, referred to as abnormal state A) and the observed state is high, it can be said that the observed state is close to abnormal state A. Conversely, if the relevance between abnormal state A and the observed state is low, it can be said that the observed state is far from abnormal state A.

[0013] The detection threshold determination unit determines a single detection threshold for detecting the degree of abnormality of the production facility based on a plurality of abnormal thresholds and relevance. The plurality of abnormal thresholds are thresholds of abnormal indicators corresponding to each of the plurality of abnormal states. For example, focusing on a certain abnormal threshold and the corresponding abnormal state (here, referred to as abnormal state A), when the abnormal indicator exceeds, for example, 80% of the abnormal threshold, it can be determined that the production facility is in abnormal state A. In the present disclosure, such a plurality of abnormal thresholds are used, and a single detection threshold is determined from the plurality of abnormal thresholds and the relevance. For example, it is conceivable to determine a single detection threshold so that the influence of those with relatively high relevance among the plurality of abnormal thresholds becomes greater. In this way, in the present disclosure, a single detection threshold is determined in consideration of a plurality of abnormal states that may occur in the production facility comprehensively.

[0014] The abnormality detection unit detects the degree of abnormality of the production facility based on an abnormality index and a detection threshold. The degree of abnormality of the production facility refers to the degree indicating how much the production facility deviates from the normal state as a whole. If this degree of abnormality becomes too high, the production facility may become unable to operate without maintenance. Therefore, it is very important for the person managing the production facility to know this degree of abnormality. In the present disclosure, in addition to the abnormality index that can be used alone for detecting the abnormality of the production facility as described above, the degree of abnormality of the production facility is detected using a single detection threshold determined in consideration of a plurality of abnormal states. That is, in the present disclosure, equipment abnormalities caused by a plurality of factors can be detected by using only the abnormality index and a single detection threshold. The person managing the production facility can perform maintenance on the production facility at an appropriate timing based on the degree of abnormality or the information on equipment abnormalities detected in such a simple manner. Furthermore, since the person managing the production facility can also know the information on the degree of relevance, when performing maintenance, the maintenance target can be appropriately selected based on the degree of relevance.

[0015] The feature amount may include a first feature amount and a second feature amount. The production facility monitoring system may further include a normal model that models the normal state of the production facility, with the first feature amount obtained from the facility information of the production facility in the normal state as the input and the second feature amount obtained from the facility information of the production facility in the normal state as the output. The abnormal index determination unit may determine an abnormal index of the production facility based on the second feature amount output from the normal model with the first feature amount of the observed facility information as the input and the second feature amount of the observed facility information. According to this configuration, the abnormal index determination unit can determine the abnormal index of the production facility by utilizing a predetermined normal model. As the determination method, for example, it is conceivable that the greater the difference between the second feature amount output from the normal model with the first feature amount of the observed facility information as the input and the second feature amount of the observed facility information, the greater the abnormal index of the production facility. Note that, as the first feature amount, for example, in the case of a plasma processing apparatus such as a plasma cleaner or a plasma dicer, it may be information regarding a recipe for executing a process, and such a recipe may include control information. In this case, as the second feature amount, for example, it may be information output from the production facility corresponding to the recipe for executing the process, or it may be sensor information.

[0016] Taking the abnormal state with the highest degree of relevance to the observation state as the first abnormal state, and the abnormal state with the second highest degree of relevance to the observation state as the second abnormal state, the detection threshold determination unit may determine the detection threshold by interpolating the abnormal threshold of the first abnormal state and the abnormal threshold of the second abnormal state based on the Mahalanobis distance between the first abnormal state and the observation state, and the Mahalanobis distance between the second abnormal state and the observation state. According to this configuration, a single detection threshold is determined taking into account the first abnormal state and the second abnormal state, both of which have a high degree of relevance to the observation state. Here, in addition to the first abnormal state and the second abnormal state, there may be many abnormal states in the production equipment. However, when an equipment abnormality actually occurs in the production equipment, it is often sufficient to consider the top two abnormal states, that is, the first abnormal state and the second abnormal state. In this way, instead of comprehensively processing information on countless abnormal states, by processing information on the first abnormal state and the second abnormal state, that is, a necessary and sufficient amount of information, equipment abnormalities can be detected more simply.

[0017] The abnormality degree detection unit may further include a notification unit that notifies the abnormality degree of the production equipment. The notification unit can notify the abnormality degree by any method. For example, the notification unit may notify the abnormality degree by a visual method such as a display, or alternatively, by an auditory method or a tactile method. The person managing the production equipment can appropriately determine the maintenance time of the production equipment based on the notified abnormality degree information.

[0018] When notifying the abnormality degree of the production equipment, the notification unit may notify the abnormal state with the highest degree of relevance to the observation state. By knowing such an abnormal state, the person managing the production equipment can directly know the location where the need for maintenance is the highest. Therefore, the person managing the production equipment can easily select the maintenance target.

[0019] When the notification unit notifies the degree of abnormality of the production equipment, it may also notify the abnormal state with the second highest degree of relevance to the observation state. By knowing such an abnormal state, the person in charge of managing the production equipment can directly know the location where the need for maintenance is the second highest. Therefore, the person in charge of managing the production equipment can add such a location to the maintenance target as needed.

[0020] The production equipment may be a plasma processing apparatus including a processing chamber where plasma processing is performed, and the degree of abnormality of the production equipment may include the degree of abnormality of elements arranged in the processing chamber. In such a plasma processing apparatus, there is a situation that the opening frequency of the processing chamber is desired to be as low as possible in terms of the operation management of the apparatus. According to the present disclosure, based on two indexes, namely, the degree of abnormality of elements in the processing chamber and the degree of relevance in the production equipment, it is possible to predict the timing when maintenance of each element is required. The person in charge of managing the plasma processing apparatus can also perform maintenance on a plurality of elements in the processing chamber collectively based on such a prediction. In that case, for example, the opening frequency of the processing chamber can be reduced compared to the case where maintenance is performed each time an abnormality occurs in each element.

[0021] (Production Equipment Monitoring Method) The production equipment monitoring method according to the present disclosure includes an abnormal index determination step, a relevance determination step, a detection threshold determination step, and an abnormality degree detection step.

[0022] In the abnormal index determination step, an abnormal index of the production equipment is determined based on a feature amount obtained from the equipment information of the production equipment. The equipment information may be, for example, sensor information or control information. The feature amount may be obtained by performing extraction or calculation according to a predetermined rule based on such equipment information. Although it is possible to directly detect an abnormality of the production equipment from the abnormal index, in the present disclosure, the abnormality of the production equipment is not detected only using the abnormal index.

[0023] In the relevance determination step, the relevance between each of a plurality of abnormal states that may occur in the production equipment and the observed state of the production equipment is determined. If the relevance between a certain abnormal state (here, referred to as abnormal state A) and the observed state is high, it can be said that the observed state is close to abnormal state A. Conversely, if the relevance between abnormal state A and the observed state is low, it can be said that the observed state is far from abnormal state A.

[0024] In the detection threshold determination step, a single detection threshold for detecting the degree of abnormality of the production equipment is determined based on a plurality of abnormality thresholds and relevance. For example, it is conceivable to determine a single detection threshold so that the influence of those with relatively high relevance among the plurality of abnormality thresholds becomes greater. In this way, in the present disclosure, a single detection threshold is determined in consideration of a plurality of abnormal states that may occur in the production equipment.

[0025] In the abnormality degree detection step, the abnormality degree of the production equipment is detected based on the abnormality index and the detection threshold. If this abnormality degree becomes too high, the production equipment may become a state where it cannot operate without maintenance. Therefore, for those who manage the production equipment, it is very important to know this abnormality degree. In the present disclosure, in addition to the abnormality index that can be used alone for detecting abnormalities in the production equipment as described above, the abnormality degree of the production equipment is detected using a single detection threshold determined in consideration of a plurality of abnormal states. That is, in the present disclosure, equipment abnormalities caused by a plurality of factors can be detected by using only the abnormality index and a single detection threshold. Those who manage the production equipment can maintain the production equipment at an appropriate timing based on the abnormality degree or equipment abnormality information detected in such a simple manner. Furthermore, since those who manage the production equipment can also know the relevance information, they can appropriately select the maintenance target based on the relevance when performing maintenance.

[0026] The feature amount may include a first feature amount and a second feature amount. In the abnormal index determination step, the first feature amount obtained from the equipment information of the production equipment in the normal state is used as an input, and the second feature amount obtained from the equipment information of the production equipment in the normal state is used as an output to model the normal state of the production equipment. Based on the second feature amount output after inputting the first feature amount of the observed equipment information into the normal model, and the second feature amount of the observed equipment information, the abnormal index of the production equipment may be determined. According to this configuration, in the abnormal index determination step, the abnormal index of the production equipment can be determined by utilizing a predetermined normal model. As a determination method, for example, the larger the difference between the second feature amount output from the normal model with the first feature amount of the observed equipment information as an input and the second feature amount of the observed equipment information, the larger the abnormal index of the production equipment is considered to be.

[0027] As described above, according to the present disclosure, by using an abnormal index and a single detection threshold value, equipment abnormalities caused by a plurality of factors can be easily detected. Furthermore, according to the present disclosure, a person who manages the production equipment can perform maintenance on the production equipment at an appropriate timing and after appropriately selecting the maintenance target.

[0028] Hereinafter, an example of the production equipment monitoring system and the production equipment monitoring method according to the present disclosure will be specifically described with reference to the drawings. The components and steps of the example of the production equipment monitoring system and the production equipment monitoring method described below can apply the components and steps described above. The components and steps of the example of the production equipment monitoring system and the production equipment monitoring method described below can be changed based on the above description. Also, the matters described below may be applied to the above embodiments. Among the components and steps of the example of the production equipment monitoring system and the production equipment monitoring method described below, the components and steps that are not essential for the production equipment monitoring system and the production equipment monitoring method according to the present disclosure may be omitted. Note that the drawings shown below are schematic and do not accurately reflect the shapes and numbers of actual members.

[0029] As shown in Fig. 1, the production facility of this embodiment is a plasma processing apparatus 10 (specifically, a plasma cleaner) including a processing chamber where plasma processing of a substrate 1 (object) is performed. Note that the plasma processing apparatus 10 may be a plasma dicing machine for performing plasma dicing. Also, the production facility may be a production facility other than the plasma processing apparatus.

[0030] Hereinafter, the configurations of the plasma processing apparatus 10 and the production facility monitoring system 20 will be described, and then the production facility monitoring method will be described.

[0031] (Plasma Processing Apparatus) The plasma processing apparatus 10 includes a base 11, a lid 12, and a power supply unit 13. In the plasma processing apparatus 10, the processing chamber is defined by the base 11 and the lid 12.

[0032] The base 11 has a base body 11a, an electrode body 11b supported by the base body 11a and facing the lid 12, and a guide 11c disposed on the electrode body 11b. The base body 11a is a rectangular frame-shaped member and is electrically insulated from the electrode body 11b. The electrode body 11b functions as one electrode when generating plasma in the processing chamber. The guide 11c is composed of at least a pair of rail-shaped members extending in a predetermined direction (the left-right direction in Fig. 1). The guide 11c guides the substrate 1 along the predetermined direction. The guide 11c is an example of an element disposed in the processing chamber.

[0033] The lid 12 is box-shaped with a ceiling portion and side walls extending from around the ceiling portion. The lid 12 is openable and closable by an opening / closing mechanism (not shown). The side walls of the lid 12 are in close contact with the peripheral edge of the base body 11a when the lid 12 is closed. Thereby, a processing chamber is formed inside the lid 12 and the base 11. When the lid 12 opens, the processing chamber is opened. On the lower surface of the ceiling portion of the lid 12, an irradiation portion 12a for irradiating plasma is formed. The irradiation portion 12a functions as the other electrode when generating plasma in the processing chamber. On the inner surface of the side wall of the lid 12, a plasma monitor 12b for detecting the light intensity during plasma generation is provided. The irradiation portion 12a and the plasma monitor 12b are each an example of an element disposed inside the processing chamber.

[0034] The power supply unit 13 has a high-frequency power supply 13a and an automatic matcher 13b. The high-frequency power supply 13a is electrically connected to the electrode body 11b of the base 11. The high-frequency power supply 13a applies high-frequency power between the electrode body 11b and the lid 12 in a state where a process gas exists in the processing chamber. Thereby, plasma is generated in the processing chamber. The automatic matcher 13b has a function of preventing interference caused by the reflected wave of the high-frequency applied between the electrode body 11b and the lid 12.

[0035] Although not shown in the figure, the plasma processing apparatus 10 further includes a control unit and various sensors. The control unit controls the opening and closing of the lid 12, the operation of the power supply unit 13, and the supply operation of the process gas to the processing chamber. The control unit includes a CPU and a storage device storing programs executable by the CPU. The various sensors detect various states of the plasma processing apparatus 10. Examples of the types of sensors include a sensor for detecting the gas pressure inside the processing chamber and a sensor for detecting the output signal of the plasma monitor 12b. The information of the signals of the control unit (control information) and the information of the signals of the various sensors (sensor information) are used by the production facility monitoring system 20 as described later.

[0036] (Production Facility Monitoring System) As shown in FIG. 2, the production facility monitoring system 20 is a system for monitoring the plasma processing apparatus 10. The production facility monitoring system 20 is composed of a computer communicably connected to the plasma processing apparatus 10. The production facility monitoring system 20 includes a feature quantity generation unit 21, a normal model 22, an abnormality index determination unit 23, a relevance determination unit 24, a detection threshold determination unit 25, an abnormality degree detection unit 26, and a notification unit 27.

[0037] The feature quantity generation unit 21 generates a first feature quantity and a second feature quantity based on the control information acquired from the control unit of the plasma processing apparatus 10 and the sensor information acquired from various sensors of the plasma processing apparatus 10. As methods for generating the first feature quantity and the second feature quantity from the control information and the sensor information, various methods can be adopted. For example, predetermined information may be extracted from the control information and the sensor information as each feature quantity, or alternatively or additionally, various operations may be performed on the information extracted from the control information and the sensor information to generate each feature quantity.

[0038] The normal model 22 is a learned model that models the normal state of the plasma processing apparatus 10. The normal model 22 may be generated, for example, by setting a multiple regression distribution model that takes the first feature quantity of the plasma processing apparatus 10 in the normal state as an input and the second feature quantity of the plasma processing apparatus 10 in the normal state as an output, and learning the convergence point of the multiple regression distribution model by machine learning. When the first feature quantity of the equipment information (control information and sensor information) observed during the operation of the plasma processing apparatus 10 is input to the normal model 22, the corresponding second feature quantity is output.

[0039] The abnormality index determination unit 23 determines an abnormality index a of the plasma processing apparatus 10 based on the second feature amount output by the normal model 22 as described above and the second feature amount of the facility information observed during the operation of the plasma processing apparatus 10. For example, the abnormality index determination unit 23 may determine a single abnormality index a based on the following formula (1). In formula (1), x’ is the feature amount of the observed value (second feature amount), μ^ is the average value of each feature amount (each second feature amount), and σ^ is the square root of the variance of the feature amount (second feature amount).

[0040]

Equation

[0041] The relevance determination unit 24 determines the relevance D between each of a plurality of abnormal states that can occur in the plasma processing apparatus 10 and the observed state of the plasma processing apparatus 10. The plurality of abnormal states include, for example, a state where the irradiation unit 12a is dirty, a state where the area between the base 11 and the lid 12 is dirty, a state where the plasma monitor 12b is dirty, and a state where the operation of the power supply unit 13 is defective. The relevance determination unit 24 may, for example, obtain the Mahalanobis distance between the second feature amount of the observed facility information and the data distribution of each abnormal state, and determine the relevance D by taking the reciprocal of the Mahalanobis distance and normalizing them. Here, the data distribution of each abnormal state may be created, for example, based on the feature amount of the facility information obtained in a state where an abnormal state is intentionally generated in the plasma processing apparatus 10. For example, based on the feature amount of the facility information obtained in a state where dirt is attached to the irradiation unit 12a (or a state where a member simulating dirt is attached), a data distribution corresponding to the abnormal state where the irradiation unit 12a is dirty can be generated.

[0042] The detection threshold determination unit 25 determines a single detection threshold th for detecting the degree of abnormality A of the plasma processing apparatus 10 based on a plurality of abnormality thresholds at and relevance degrees D. The abnormality threshold at is a threshold of the abnormality index a corresponding to each abnormal state. For example, the abnormality threshold at of the irradiation unit 12a is a threshold of the abnormality index a corresponding to the abnormal state of the irradiation unit 12a. The magnitudes of the plurality of abnormality thresholds at are usually different from each other. For example, it is assumed that the abnormality threshold at corresponding to the abnormal state of the irradiation unit 12a is 100, while the abnormality threshold at corresponding to the abnormal state of the plasma monitor 12b is 50. In this case, even if the abnormality index a is simply compared with each abnormality threshold at, it is not possible to appropriately determine the abnormality of the entire plasma processing apparatus 10. For example, assuming the numerical values exemplified in this paragraph, even if a value of 50 is obtained as the abnormality index a, although it has reached the abnormality threshold at of 50 of the plasma monitor 12b, there is a large margin with respect to the abnormality threshold at of 100 of the irradiation unit 12a, so it cannot be determined whether the abnormality index a of 50 is a problem for the entire plasma processing apparatus 10.

[0043] On the other hand, the detection threshold determination unit 25 determines a single detection threshold th as described above. In the present embodiment, the abnormal state with the highest relevance degree D to the observed state of the plasma processing apparatus 10 is set as the first abnormal state, and the abnormal state with the second highest relevance degree D to the observed state is set as the second abnormal state. Then, the detection threshold determination unit 25 determines a single detection threshold th by interpolating the abnormality threshold at of the first abnormal state and the abnormality threshold at of the second abnormal state based on the Mahalanobis distance between the first abnormal state and the observed state and the Mahalanobis distance between the second abnormal state and the observed state.

[0044] For example, the abnormal state of the irradiation unit 12a (abnormal threshold at = 100) is set as the first abnormal state, and the abnormal state of the plasma monitor 12b (abnormal threshold at = 50) is set as the second abnormal state. Also, the Mahalanobis distance between the observed state of the plasma processing apparatus 10 and the first abnormal state is set to 1, and the Mahalanobis distance between the observed state and the second abnormal state is set to 4. In this case, if the abnormal thresholds at of the first abnormal state and the second abnormal state are linearly interpolated, the single detection threshold th will be 90. This value is obtained by solving the equation (100 - th):(th - 50) = 1:4. However, the interpolation method is not limited to linear interpolation, and interpolation can be performed using any function.

[0045] The abnormality degree detection unit 26 detects the abnormality degree A of the plasma processing apparatus 10 based on the abnormality index a and the detection threshold th. For example, the abnormality degree detection unit 26 may detect the abnormality degree A as the ratio of the abnormality index a to the detection threshold th (A = a / th). As a specific example, when the abnormality index a is 50 and the detection threshold th is 90, the abnormality degree A is approximately 0.56 (= 50 / 90). Thus, in the production facility monitoring system 20 of the present embodiment, the overall abnormality degree A of the plasma processing apparatus 10 can be easily detected based on the comparison between one abnormality index a and one detection threshold th.

[0046] The notification unit 27 notifies the abnormality degree A of the plasma processing apparatus 10. The notification unit 27 of the present embodiment is a computer display, but is not limited thereto. The notification unit 27 may notify the abnormality degree A, for example, by displaying the value of the abnormality degree A on the display. Also, when notifying the abnormality degree A, the notification unit 27 notifies the abnormal state (first abnormal state) with the highest degree of relevance D to the observed state of the plasma processing apparatus 10 and the abnormal state (second abnormal state) with the second highest degree of relevance D. That is, the notification unit 27 of the present embodiment notifies not only the abnormality degree A of the entire plasma processing apparatus 10 but also the two factors with high contribution degrees to the abnormality degree A. Therefore, the person managing the plasma processing apparatus 10 can appropriately determine the maintenance timing and maintenance target of the plasma processing apparatus 10 before a failure occurs.

[0047] (Production Equipment Monitoring Method) Next, the production equipment monitoring method of this embodiment will be described. The production equipment monitoring method may be executed in the above-described production equipment monitoring system 20, or may be executed in a system having other configurations.

[0048] As shown in FIG. 3, the production equipment monitoring method includes a feature quantity generation step S1, an abnormality index determination step S2, a relevance determination step S3, a detection threshold determination step S4, an abnormality degree detection step S5, and a notification step S6.

[0049] In the feature quantity generation step S1, a first feature quantity and a second feature quantity are generated based on the control information acquired from the control unit of the plasma processing apparatus 10 and the sensor information acquired from various sensors of the plasma processing apparatus 10. As a method for generating the first feature quantity and the second feature quantity from the control information and the sensor information, various methods can be adopted. For example, predetermined information may be extracted from the control information and the sensor information as each feature quantity, or instead of or in addition to this, various operations may be performed on the information extracted from the control information and the sensor information to generate each feature quantity.

[0050] In the abnormality index determination step S2, an abnormality index a of the plasma processing apparatus 10 is determined based on the second feature quantity output by inputting the first feature quantity of the equipment information observed during the operation of the plasma processing apparatus 10 into the normal model 22 and the second feature quantity of the observed equipment information. For example, in the abnormality index determination step S2, the abnormality index a may be determined based on the above-described formula (1).

[0051] In the relevance determination step S3, a relevance D between each of a plurality of abnormal states that may occur in the plasma processing apparatus 10 and the observed state of the plasma processing apparatus 10 is determined. The relevance determination unit 24 may, for example, obtain the Mahalanobis distance between the second feature quantity of the observed equipment information and the data distribution of each abnormal state, take the reciprocal of the Mahalanobis distance, and then normalize them to determine the relevance D.

[0052] In the detection threshold determination step S4, a single detection threshold th for detecting the degree of abnormality A of the plasma processing apparatus 10 is determined based on a plurality of abnormality thresholds at and relevance degrees D. In the present embodiment, the abnormal state with the highest relevance degree D to the observed state of the plasma processing apparatus 10 is defined as the first abnormal state, and the abnormal state with the second highest relevance degree D to the observed state is defined as the second abnormal state. Then, based on the Mahalanobis distance between the first abnormal state and the observed state and the Mahalanobis distance between the second abnormal state and the observed state, a single detection threshold th is determined by interpolating the abnormality threshold at of the first abnormal state and the abnormality threshold at of the second abnormal state.

[0053] In the abnormality degree detection step S5, the abnormality degree A of the plasma processing apparatus 10 is detected based on the abnormality index a and the detection threshold th. For example, the abnormality degree detection unit 26 may detect the abnormality degree A as the ratio of the abnormality index a to the detection threshold th (A = a / th). As a specific example, when the abnormality index a is 50 and the detection threshold th is 90, the abnormality degree A is approximately 0.56 (= 50 / 90). Thus, in the production facility monitoring method of the present embodiment, the overall abnormality degree A of the plasma processing apparatus 10 can be easily detected based on the comparison between one abnormality index a and one detection threshold th.

[0054] In the notification step S6, the abnormality degree A of the plasma processing apparatus 10 is notified. In the notification step S6, for example, the abnormality degree A may be notified by displaying the value of the abnormality degree A on a display. Further, when notifying the abnormality degree A, the notification unit 27 notifies the abnormal state (first abnormal state) with the highest relevance degree D to the observed state of the plasma processing apparatus 10 and the abnormal state (second abnormal state) with the second highest relevance degree D. That is, in the notification step S6 of the present embodiment, not only the abnormality degree A of the entire plasma processing apparatus 10 but also two factors with high contribution degrees to the abnormality degree A are notified. Therefore, the person managing the plasma processing apparatus 10 can appropriately judge the maintenance timing and maintenance target of the plasma processing apparatus 10 before a failure occurs.

Industrial Applicability

[0055] The present disclosure can be used in a production facility monitoring system and a production facility monitoring method.

Explanation of Signs

[0056] 1: Substrate 10: Plasma processing apparatus (production facility) 11: Base 11a: Substrate body 11b: Electrode body 11c: Guide (element) 12: Lid 12a: Irradiation part (element) 12b: Plasma monitor (element) 13: Power supply unit 13a: High-frequency power supply 13b: Automatic matcher 20: Production facility monitoring system 21: Feature quantity generation unit 22: Normal model 23: Abnormality index determination unit 24: Relevance determination unit 25: Detection threshold determination unit 26: Abnormality degree detection unit 27: Notification unit a: Abnormality index at: Abnormality threshold A: Abnormality degree D: Relevance th: Detection threshold

Claims

1. An abnormality index determination unit that determines an abnormality index of the production equipment based on a feature amount obtained from equipment information of the production equipment; A relevance determination unit that determines a relevance between each of a plurality of abnormal states that may occur in the production equipment and an observed state of the production equipment; A detection threshold determination unit that determines a single detection threshold for detecting an abnormality degree of the production equipment based on a plurality of abnormality thresholds that are thresholds of the abnormality index corresponding to each of the plurality of abnormal states and the relevance; An abnormality degree detection unit that detects an abnormality degree of the production equipment based on the abnormality index and the detection threshold; Comprising; The equipment information includes control information that is information of a control signal used for controlling the production equipment, and sensor information that is information of a signal output by a sensor provided in the production equipment corresponding to the control information; The feature amount includes a first feature amount obtained from the control information and a second feature amount obtained from the sensor information; Further comprising a normal model that models a normal state of the production equipment, and outputs the second feature amount obtained from the sensor information of the production equipment in the normal state when the first feature amount obtained from the control information of the production equipment in the normal state is input; The abnormality index determination unit determines the abnormality index of the production equipment based on the second feature amount output by inputting the first feature amount obtained from the equipment information observed during operation of the production equipment into the normal model, and the second feature amount obtained from the equipment information observed during operation; Regarding the abnormal state with the highest relevance to the observed state as the first abnormal state, and regarding the abnormal state with the second highest relevance to the observed state as the second abnormal state; The detection threshold determination unit determines the detection threshold by interpolating the abnormality threshold of the first abnormal state and the abnormality threshold of the second abnormal state based on the Mahalanobis distance between the first abnormal state and the observed state, and the Mahalanobis distance between the second abnormal state and the observed state; A production equipment monitoring system.

2. The production equipment monitoring system according to claim 1, wherein the abnormality degree detection unit further comprises a notification unit that notifies an abnormality degree of the production equipment.

3. The production equipment monitoring system according to claim 2, wherein the notification unit notifies the abnormal state with the highest relevance to the observed state when notifying the abnormality degree of the production equipment.

4. The production equipment monitoring system according to claim 3, wherein when the notification unit notifies the abnormality degree of the production equipment, the notification unit also notifies the abnormality state having the second highest degree of relevance with the observation state.

5. The production equipment is a plasma processing apparatus including a processing chamber in which plasma processing is performed. The production equipment monitoring system according to any one of claims 1 to 4, wherein the abnormality degree of the production equipment includes the abnormality degree of elements arranged in the processing chamber.

6. An abnormality index determination step of determining an abnormality index of the production equipment based on a feature amount obtained from equipment information of the production equipment; A relevance determination step of determining the relevance between each of a plurality of abnormality states that can occur in the production equipment and the observation state of the production equipment; A detection threshold determination step of determining a single detection threshold for detecting the abnormality degree of the production equipment based on a plurality of abnormality thresholds that are thresholds of the abnormality indexes corresponding to each of the plurality of abnormality states and the relevance; An abnormality degree detection step of detecting the abnormality degree of the production equipment based on the abnormality index and the detection threshold; comprising The equipment information includes control information that is information of a control signal used for controlling the production equipment, and sensor information that is information of a signal output by a sensor provided in the production equipment corresponding to the control information. The feature amount includes a first feature amount obtained from the control information and a second feature amount obtained from the sensor information. A normal model that models a normal state of the production equipment, wherein when the first feature amount obtained from the control information of the production equipment in the normal state is input, the second feature amount obtained from the sensor information of the production equipment in the normal state is output. In the abnormality index determination step, the abnormality index of the production equipment is determined based on the second feature amount output by inputting the first feature amount obtained from the equipment information observed during operation of the production equipment into the normal model and the second feature amount obtained from the equipment information observed during operation. The abnormality state having the highest degree of relevance with the observation state is defined as a first abnormality state, and the abnormality state having the second highest degree of relevance with the observation state is defined as a second abnormality state. The detection threshold determination step determines the detection threshold by interpolating the anomaly threshold of the first abnormal state and the anomaly threshold of the second abnormal state based on the Mahalanobis distance between the first abnormal state and the observation state and the Mahalanobis distance between the second abnormal state and the observation state. Production equipment monitoring method.

Images