JP2020139886A - Substrate inspection device, substrate inspection method, and program - Google Patents

Abstract

To provide a technology for quick and accurate substrate product inspection.SOLUTION: A substrate inspection device 10 includes an image acquisition part 110 and an abnormality determination part 120. The image acquisition part 110 acquires a plurality of images generated by capturing a substrate under a different condition. The abnormality determination part 120 determines presence of abnormality on the substrate by processing the images acquired by the image acquisition part 110 using a discriminator 122 generated by machine learning.SELECTED DRAWING: Figure 1

Description

The present invention relates to a substrate inspection apparatus, a substrate inspection method, and a program.

The substrate manufacturing process includes an inspection step for detecting defects in the manufactured substrate. For example, in Patent Document 1 below, a portion necessary for inspection is cut out from an image (image of a substrate) captured by an inspection image imaging unit to obtain an inspection image, and the inspection image is compared with a reference image or an adjacent image. By doing so, it is disclosed that the substrate is inspected.

Japanese Unexamined Patent Publication No. 2015-190826

At the site where the substrate is manufactured, it is required to realize quick and accurate inspection of the substrate product.

The present invention has been made in view of the above-mentioned problems. One of the objects of the present invention is to provide a technique for realizing a quick and accurate inspection of a substrate product.

According to the present invention
An image acquisition unit that acquires multiple images generated by shooting the board under different conditions,
By processing the plurality of images using a classifier generated by machine learning,
An abnormality determination unit that determines the presence or absence of an abnormality on the substrate,
A substrate inspection device is provided.

According to the present invention
The computer
Acquire multiple images generated by shooting the board under different conditions,
By processing the plurality of images using a classifier generated by machine learning,
Judging whether or not there is an abnormality in the substrate,
A substrate inspection method including the above is provided.

According to the present invention, a program for causing a computer to execute the above-mentioned substrate inspection method is provided.

According to the present invention, it is possible to provide a technique for realizing a quick and accurate inspection of a substrate product.

It is a figure which illustrates the functional structure of the substrate inspection apparatus in 1st Embodiment. It is a block diagram which illustrates the hardware structure of the board inspection apparatus. It is a flowchart which illustrates the flow of the process executed by the substrate inspection apparatus of 1st Embodiment. It is a figure which shows the functional structure of the substrate inspection apparatus which concerns on 2nd Embodiment. It is a flowchart which illustrates the flow of the process executed by the substrate inspection apparatus of 2nd Embodiment. It is a figure which shows an example of the information used when deciding the post-processing to be performed on the inspection target substrate which was judged to have an abnormality. It is a figure which shows an example of the information used when deciding the post-processing to be performed on the inspection target substrate which was judged to have abnormality. It is a figure which illustrates the functional structure of the substrate inspection apparatus which concerns on 3rd Embodiment. It is a flowchart which shows the flow of the process executed by the substrate inspection apparatus of 3rd Embodiment. It is a figure which shows an example of the information which associates the classifier and the attribute corresponding to the classifier. It is a flowchart which illustrates the flow of the process executed by the substrate inspection apparatus of the modification of 1st Embodiment.

The present invention covers all substrate products including ceramic substrates and ceramic circuit boards. Here, the ceramic substrate is a ceramic plate produced by forming and sintering a raw material powder. A ceramic circuit board is a composite (ceramic metal composite) in which the above-mentioned ceramic substrate and a metal plate (for example, a copper plate or an aluminum plate) are joined by using a brazing material or the like, and resist coating, etching, or plating is performed on the composite. Therefore, a desired circuit pattern is formed.

Hereinafter, embodiments of the present invention will be described with reference to the drawings, but the present invention should not be construed as being limited to these, and is based on the knowledge of those skilled in the art as long as it does not deviate from the gist of the present invention. Therefore, various changes and improvements can be made. The plurality of components disclosed in the embodiments can form various inventions by appropriate combinations. For example, some components may be deleted from all the components shown in the embodiment, or components of different embodiments may be combined as appropriate.

[First Embodiment]
<Function configuration example>
FIG. 1 is a diagram illustrating a functional configuration of a substrate inspection device according to the first embodiment. As shown in FIG. 1, the substrate inspection device 10 of the present embodiment includes an image acquisition unit 110 and an abnormality determination unit 120.

The image acquisition unit 110 acquires a plurality of images generated by photographing the inspection target substrate under different conditions. The image of the substrate to be inspected is generated by using an imaging device (not shown). As an example, by photographing the inspection target substrate from various angles using a plurality of imaging devices, a plurality of images of the inspection target substrate are generated. As another example, the inspection target is examined by photographing the inspection target substrate multiple times using one or more imaging devices while changing the characteristics of the light applied to the inspection target substrate (wavelength of light, amount of light, etc.). Multiple images of the substrate are generated. The image acquisition unit 110 acquires a part or all of the generated plurality of images.

The abnormality determination unit 120 determines the presence or absence of an abnormality in the substrate to be inspected by processing a plurality of images acquired by the image acquisition unit 110 using the classifier 122 generated by machine learning. Further, the abnormality determination unit 120 may be further configured to classify the inspection target substrate determined to have an abnormality into a plurality of groups. The following is an example in which the abnormality determination unit 120 determines whether or not there is an abnormality in the inspection target substrate, and further classifies the inspection target substrate determined to have an abnormality into a plurality of groups based on the abnormality. Explain to. In the following description, the "inspection target substrate determined to have an abnormality" is also referred to as "abnormal substrate".

Here, the classifier 122 is a learning model constructed so that an abnormality on an inspection target substrate can be detected by using a known machine learning algorithm (for example, deep learning or SVM (Support Vector Machine)). For example, by performing learning using a data set for learning (a combination of image data of anomalies that can occur on the board and information that identifies the type of the anomaly), anomalies can be found on the substrate from the feature amount of the image. A classifier 122 is constructed that can determine whether or not it exists and, if an abnormality exists on the substrate, what kind of abnormality it is. The machine learning algorithm used to construct the classifier 122 is not particularly limited as long as it can identify or classify various abnormalities of the substrate to be inspected. When the classifier 122 receives the image of the inspection target substrate acquired by the image acquisition unit 110 as an input, the discriminator 122 receives the image of the inspection target substrate acquired by the image acquisition unit 110 as an input, and based on the model constructed by machine learning, the inspection target substrate shown in the image has an abnormality (a substrate ( It is determined whether or not the substrate has an abnormality. Further, when the discriminator 122 determines that the substrate to be inspected has an abnormality, the classifier 122 further outputs information indicating the type of abnormality possessed by the substrate. Based on the information output from the classifier 122, the abnormality determination unit 120 determines whether or not the inspection target substrate is an abnormality-existing substrate, and when it is determined that the inspection target substrate is an abnormality-existing substrate, the abnormality determination unit 120 has an abnormality. The type can be identified.

As an example, when the substrate to be inspected is a ceramic substrate, the classifier 122 indicates that the input image of the substrate to be inspected has, for example, cracks generated in the ceramic substrate, dirt or foreign matter on the substrate surface, scratches, chips, or It is constructed so as to detect unevenness as "abnormality". In this case, the abnormality determining unit 120 determines as a substrate abnormality a crack generated in the ceramic substrate, the presence of dirt or foreign matter on the substrate surface, scratches, chips, or irregularities based on the information output from the classifier 122. be able to. As another example, when the substrate to be inspected is a circuit board, the classifier 122 regards the input image of the substrate to be inspected as "abnormal", for example, the presence of dirt or foreign matter on the substrate surface, scratches, or irregularities. Constructed to detect. In this case, the abnormality determination unit 120 determines cracks generated in the ceramic substrate, at least the presence of dirt or foreign matter on the substrate surface, scratches, or irregularities as abnormalities on the substrate, based on the information output from the classifier 122. can do. Further, the classifier 122 has an abnormality that actually exists on the substrate as described above, and an abnormality that does not actually exist on the substrate (for example, an abnormality that occurs only on the image data due to the environment at the time of shooting such as halation or the structure of the imaging device. ) May be identifiable.

Here, the handling (post-processing) of the substrate differs depending on the type of abnormality that the substrate with abnormality has. For example, if the output information of the classifier 122 includes only "abnormality that does not exist on the board" as the type of abnormality for the board to be inspected that is determined to have an abnormality, there is no substantial abnormality in the board. I can say. In this case, the substrate does not require any special post-treatment and can be shipped as an acceptable product as it is. Further, for example, when the output information of the classifier 122 includes "abnormality existing in the board" as the type of abnormality for the board to be inspected determined to be a board with an abnormality, some post-processing is required for the board. It becomes. Then, the post-processing to be performed on the substrate changes according to the abnormality that the substrate has. For example, if the abnormality of the substrate to be inspected determined to be an abnormal substrate is dirt or foreign matter adhering to the surface of the substrate, the substrate can be shipped as a accepted product after undergoing a predetermined cleaning treatment. .. Further, for example, when the abnormality of the substrate to be inspected determined as the substrate with abnormality is a slight scratch or the like, the substrate can be shipped as a accepted product after undergoing a predetermined repair process. Further, for example, when the abnormality of the inspection target substrate determined as the substrate with abnormality is a crack or a scratch so severe that there is no possibility of repair, the substrate is disposed of, for example.

As described above, the inspection target substrate determined to have an abnormality is handled differently depending on the type of abnormality possessed by the substrate. Therefore, the abnormality determination unit 120 determines the group to which the substrate belongs (that is, the group classified based on the type of abnormality) based on the type of abnormality possessed by the substrate determined as the substrate with abnormality. For example, as a result of inputting an image of the substrate to be inspected into the classifier 122, it is determined that the substrate to be inspected has an abnormality, and the types of the abnormality that the substrate has are "abnormality that does not actually exist on the substrate" and "dirt on the substrate surface". It is assumed that an output indicating "abnormality due to a foreign substance" is obtained. In this case, the abnormality determination unit 120 is based on the output of the classifier 122, and is a group corresponding to "an abnormality in which the substrate with abnormality does not actually exist in the substrate" and a group corresponding to "an abnormality due to dirt or foreign matter on the substrate surface". Judged to belong to.

[Hardware configuration example]
Each functional component of the substrate inspection device 10 may be realized by hardware that realizes each functional component (eg, a hard-wired electronic circuit, etc.), or a combination of hardware and software (eg, example). It may be realized by a combination of an electronic circuit and a program that controls it). Hereinafter, a case where each functional component of the substrate inspection device 10 is realized by a combination of hardware and software will be further described.

FIG. 2 is a block diagram illustrating a hardware configuration of the substrate inspection device 10. The hardware configuration of the substrate inspection device 10 shown in FIG. 2 is merely an example, and the hardware configuration of the substrate inspection device 10 is not limited to the configuration of FIG.

As shown in FIG. 2, the board inspection device 10 includes a bus 1010, a processor 1020, a memory 1030, a storage device 1040, an input / output interface 1050, and a network interface 1060.

The bus 1010 is a data transmission line for the processor 1020, the memory 1030, the storage device 1040, the input / output interface 1050, and the network interface 1060 to transmit and receive data to and from each other. However, the method of connecting the processors 1020 and the like to each other is not limited to the bus connection.

The processor 1020 is a processor realized by a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or the like.

The memory 1030 is a main storage device realized by a RAM (Random Access Memory) or the like.

The storage device 1040 is an auxiliary storage device realized by an HDD (Hard Disk Drive), an SSD (Solid State Drive), a memory card, a ROM (Read Only Memory), or the like. The storage device 1040 stores a program module that realizes each function of the substrate inspection device 10 (image acquisition unit 110, abnormality determination unit 120, etc.). When the processor 1020 reads each of these program modules into the memory 1030 and executes them, each function corresponding to the program module is realized.

The input / output interface 1050 is an interface for connecting the board inspection device 10 and peripheral devices. In the example of FIG. 2, the image pickup device 1052 and the input / output device 1054 are connected to the substrate inspection device 10 via the input / output interface 1050. The image pickup apparatus 1052 is an apparatus for generating an image of an inspection target substrate by using a CCD (Charge Coupled Device) image sensor, a CMOS (Complementary Metal-Oxide-Semiconductor) image sensor, or the like. As shown, one or more imaging devices 1052 are arranged so as to include the inspection target substrate mounted on the inspection table on which the inspection target substrate is placed in the imaging range. Further, the input / output device 1054 includes an image pickup device 15 for photographing the inspection target substrate, for example, an input / output device 1054 such as a mouse, a keyboard, a speaker, and a display (touch panel display).

The network interface 1060 is an interface for connecting the board inspection device 10 to the network. This network is, for example, a LAN (Local Area Network) or a WAN (Wide Area Network). The method of connecting the network interface 1060 to the network may be a wireless connection or a wired connection.

<Processing flow>
The process executed by the substrate inspection apparatus 10 of this embodiment will be described with reference to FIG. FIG. 3 is a flowchart illustrating a flow of processing executed by the substrate inspection device 10 of the first embodiment.

First, the image acquisition unit 110 acquires an image of the inspection target substrate from the image pickup apparatus 1052 connected via the input / output interface 1050 or the like (S102). The abnormality determination unit 120 inputs the image of the inspection target substrate acquired by the image acquisition unit 110 into the classifier 122 constructed in advance by machine learning (S104). Then, the abnormality determination unit 120 determines whether or not the inspection target substrate is an abnormality-existing substrate based on the output result obtained from the classifier 122 by inputting the image of the inspection target substrate ( S106). When an abnormality is detected in the inspection target substrate (S106: YES), the abnormality determination unit 120 classifies the group to be classified of the inspection target substrate (abnormality-existing substrate) determined to be abnormal into the classifier 122. Is specified based on the output result (type of abnormality) of (S108). Then, the abnormality determination unit 120 notifies a person engaged in the inspection work of the product substrate of the information about the group specified in the process of S108, for example, via a display or a lamp connected to the substrate inspection device 10 (S110). .. On the other hand, when no abnormality is detected in the inspection target substrate (S106: NO), the abnormality determination unit 120 determines the inspection target substrate as a passing product (S112). Substrates judged to be acceptable products are shipped as products after undergoing packing work.

As described above, in the present embodiment, the group of the inspection target substrate (board with abnormality) determined to have an abnormality is specified based on the type of abnormality possessed by the substrate. Then, the information about the group identified about the abnormal substrate is notified to the person engaged in the inspection work of the product substrate. By confirming the information output from the board inspection device 10 in this way, what kind of board is inspected as having an abnormality even if the person has little work experience (a person with low proficiency in work). It becomes possible to accurately and quickly determine whether or not an abnormality is present. Further, in the substrate inspection device 10 of the present embodiment, machine learning is used to detect the presence or absence of an abnormality on the product substrate. As a result, the bias in the judgment regarding the presence or absence of an abnormality in the substrate is reduced as compared with the case where a person confirms the image, and the effect of maintaining the quality of the work above a certain level can be expected regardless of the person in charge of the work.

<Modification example>
In the present embodiment, the abnormality determination unit 120 does not have to perform the process of classifying the abnormal substrate based on the abnormality. FIG. 11 is a flowchart illustrating a flow of processing executed by the substrate inspection device 10 of the modified example of the first embodiment.

The processes of S402 to S406 and S410 in FIG. 11 are the same as the processes of S102 to S106 and S110 of FIG. 3, respectively. When an abnormality is detected in the substrate to be inspected in the process of S406 (S406: YES), the abnormality determination unit 120 of this modified example inspects via an output device such as a display connected to the substrate inspection device 10. Notify the abnormality of the target board (S408). For example, the abnormality determination unit 120 indicates information indicating whether or not the substrate to be inspected has an abnormality, and the type of abnormality detected in the substrate to be inspected (dirt or foreign matter adhered, crack, scratch, chipped or uneven). Information or information indicating the detection position of an abnormality is notified via an output device.

In this modification, when the substrate inspection device 10 detects an abnormality in the substrate to be inspected in the inspection process of S20, information on the detected abnormality is notified via the output device. The information notified via the output device allows a person involved in the inspection work of the product board to judge whether or not there is an abnormality in the inspection target board, the type of abnormality detected in the inspection target board, and its detection. It will help you to grasp the position etc. Based on such information, it is expected that the post-processing of the abnormal inspection target substrate can be smoothly advanced.

[Second Embodiment]
The present embodiment is the same as the first embodiment described above, except for the points described below.

<Function configuration example>
FIG. 4 is a diagram showing a functional configuration of the substrate inspection device 10 according to the second embodiment. As shown in FIG. 4, the substrate inspection device 10 of the present embodiment further includes a post-processing identification unit 130 in addition to the configuration of the first embodiment. The post-processing identification unit 130 specifies the post-processing to be performed on the board with an abnormality by using the group classified by the abnormality determination unit 120.

Here, the post-processing to be performed on the abnormal substrate includes, for example, at least one of the processes classified as follows.
(A) Use the abnormal substrate as it is (b) Remove foreign matter adhering to the abnormal substrate to make it a product substrate (c) Repair the abnormality (scratch, warp, etc.) of the abnormal substrate and use it as the product substrate (D) Discard the abnormal substrate as a rejected product. Here, the “product substrate” means a product shipped as a accepted product.

The post-treatment of (a) above is a process of advancing the inspection target substrate determined to have an abnormality to the shipping process as it is as a passing product without performing post-treatment. When the board with anomaly is classified only into the group corresponding to "abnormality that does not actually exist on the board", the post-processing identification unit 130 performs the post-processing of the above (a) on the board with an abnormality. To determine as.

The post-treatment of (b) above is one of the treatments for eliminating the abnormality of the substrate to be inspected, and is attached to the substrate by, for example, a predetermined gas (nitrogen gas, carbon dioxide gas, etc.) or liquid (pure water, etc.). Includes cleaning and removal steps for dirt and foreign matter. When the abnormal substrate is classified into the group corresponding to "abnormality due to dirt or foreign matter on the substrate surface", the post-processing identification unit 130 should perform the post-processing of (b) above on the abnormal substrate. Determined as a process.

The post-treatment of (c) above is one of the treatments for eliminating the abnormality of the substrate to be inspected, and includes, for example, a repair step by heating or polishing. For example, sticky stains and foreign substances on the ceramic substrate, minor scratches, dry stains, etc. that cannot be removed by cleaning can be repaired by polishing treatment. Further, if the warp generated in the ceramic substrate or the like is mild, it can be repaired by heat treatment. In addition, scratches, chips, or irregularities generated on the surface of the substrate can be repaired by polishing if they are mild. When the abnormal substrate is classified into the group corresponding to "repairable abnormality occurring on the substrate surface", the post-processing identification unit 130 should execute the post-processing of (c) above on the abnormal substrate. Determined as post-processing. The post-treatment identification unit 130 includes only abnormalities such as "warp" that can be repaired by heat treatment or only abnormalities such as "scratches, chips or irregularities" that can be repaired by polishing treatment as the types of abnormalities on the substrate with abnormalities. To determine either "heat treatment" and / or "polishing treatment" as "post-treatment to be performed on the anomalous substrate" depending on whether or both of them are included. It may be configured in.

Further, as the post-treatment of (c) above, in addition to the above-mentioned heat treatment and polishing treatment, a treatment of peeling off the protective plating of the formed circuit pattern and then re-plating may be included. In this case, the classifier 122 performs machine learning using a learning data set regarding abnormalities in the plating portion such as poor adhesion of plating, variation in film thickness, rough surface of plated surface, and poor stain of plating solution residue, and plating. It is constructed so that abnormalities in the part can be detected. Then, as a result of inputting an image of the substrate to be inspected into the classifier 122, the substrate to be inspected is determined to be a substrate with an abnormality, and an output indicating "abnormality of the plated portion" is obtained as the type of abnormality possessed by the substrate. , The abnormality determination unit 120 classifies the substrate into a group corresponding to "repairable abnormality occurring on the surface of the substrate". When the post-treatment identification unit 130 classifies the abnormal substrate into a group corresponding to "repairable abnormality occurring on the substrate surface", the post-processing identification unit 130 performs a process of peeling off the protective plating and then re-plating the abnormal substrate. It is decided as the post-processing to be executed.

The post-treatment of (d) above is a treatment of disposing of the inspection target substrate, which is determined to have an abnormality, as a rejected product. When the abnormal substrate is classified into the group corresponding to the "irreparable abnormality occurring on the substrate surface", the post-processing identification unit 130 executes the post-processing of (d) above on the abnormal substrate. Determined as post-processing to be done.

Further, the post-processing to be performed on the abnormal substrate is not limited to the above-mentioned processes (a) to (d). For example, a process of reusing the abnormal substrate as a raw material for another substrate may be included as a post-processing to be performed on the abnormal substrate. The process of reusing an abnormal substrate as a raw material for another substrate is specifically a process of reusing a ceramic aggregate recovered by crushing a ceramic substrate portion as a material of another ceramic substrate. This is a process of forming a new metal plate by using the metal recovered by melting the metal part of the ceramic circuit board. When the post-processing identification unit 130 is classified into a group corresponding to "irreparable abnormality generated on the substrate surface", the post-processing identification unit 130 replaces the above-mentioned processing (d) (discarding of the substrate). , The process of reusing the abnormal substrate as a raw material for another substrate may be determined as the post-processing to be performed on the abnormal substrate.

<Processing flow>
The process executed by the substrate inspection apparatus 10 of this embodiment will be described with reference to FIG. FIG. 5 is a flowchart illustrating a flow of processing executed by the substrate inspection device 10 of the second embodiment. The process shown in FIG. 5 is executed after the process of S108 in FIG.

The post-processing specifying unit 130 determines the post-processing to be performed on the abnormal substrate based on the information about the group of the abnormal substrate identified in the process of S108 (S202).

The post-processing identification unit 130 can determine, for example, the post-processing to be performed on the inspection target substrate determined to have an abnormality by using the information shown in FIG. FIG. 6 is a diagram showing an example of information used when determining the post-processing to be performed on the inspection target substrate determined to have an abnormality. In FIG. 6, a table that defines post-processing to be executed according to the processing result of the abnormality determination unit 120 is drawn. Specifically, as the processing result of the abnormality determination unit 120, information (post-processing information) indicating post-processing to be executed is associated with each information indicating the classified group. For example, when the abnormality determination unit 120 classifies the substrate with an abnormality into group B in the processing of S108, the post-processing identification unit 130 uses the information exemplified in FIG. 6 to perform the “cleaning removal processing” for the abnormality. It can be determined as the post-processing to be performed on the existing substrate.

Then, the post-processing identification unit 130 provides information indicating post-processing to be executed on the abnormal substrate to a person engaged in the inspection work of the product substrate via, for example, a display or a lamp connected to the substrate inspection device 10. (S204). For example, the post-processing identification unit 130 causes the inspection target substrate to display a message (eg, "perform a cleaning / removing process") notifying the post-processing to be executed on the display.

As described above, in the present embodiment, the post-processing to be performed on the abnormal substrate is determined based on the classification result of the group of the abnormal substrate by the abnormality determination unit 120. Then, the information indicating the post-processing to be executed for the abnormal substrate is notified to the person engaged in the inspection work of the product substrate. A person engaged in the inspection work of the product substrate may perform what kind of post-processing on the inspection target substrate determined to have an abnormality by checking the information output from the substrate inspection device 10 of the present embodiment. Can be determined immediately. As a result, the efficiency of product substrate inspection work can be improved.

<First modification>
Here, when a plurality of types of abnormalities are present on the substrate to be inspected, the substrate may be classified into a plurality of groups depending on the types of abnormalities existing on the substrate. For example, as a result of inputting an image of the substrate to be inspected into the classifier 122, it is determined that the substrate to be inspected has an abnormality, and the types of the abnormality that the substrate has are "abnormality due to dirt or foreign matter on the substrate" and "on the surface of the substrate". It is assumed that an output indicating "a repairable abnormality has occurred" is obtained. In this case, the abnormality determination unit 120 sets the abnormal substrate into a group corresponding to "abnormality due to dirt or foreign matter on the substrate" and "repairable abnormality occurring on the substrate surface" based on the output of the classifier 122. Classify into corresponding groups. Then, the post-treatment identification unit 130 is divided into the post-treatment of (b) above corresponding to the group corresponding to "abnormality due to dirt or foreign matter on the substrate" and the group corresponding to "repairable abnormality occurring on the substrate surface". The corresponding post-processing (c) is determined as the post-processing to be performed on the abnormal substrate.

In this case, by associating each group with information indicating the priority order, the post-processing specifying unit 130 can determine the execution order of a plurality of post-processing. The post-processing specifying unit 130 can specify the post-processing execution gun station to be executed on the inspection target substrate determined to have an abnormality by using the information as shown in FIG. 7, for example. FIG. 7 is a diagram showing an example of information used when determining the post-processing to be performed on the inspection target substrate determined to have an abnormality. For example, when the substrate is classified into "Group B" and "Group C" as a result of inputting an image of the substrate to be inspected into the classifier 122, the post-processing identification unit 130 is based on the information exemplified in FIG. Therefore, the "cleaning / removing process" and the "repair process" can be determined as the post-processing to be performed on the inspection target substrate determined to have an abnormality. Further, the post-processing identification unit 130 determines the post-processing execution order based on the information indicating the priority associated with the group B and the group C among the information exemplified in FIG. 7. Can be done. In this case, the post-processing specifying unit 130 determines the order of the "cleaning / removing process" and the "repair process" as the execution order of the post-processing. Then, the post-processing identification unit 130 adds information indicating the execution order to the information indicating the post-processing to be executed for the inspection target substrate determined to have an abnormality, and adds information indicating the execution order to a person engaged in the inspection work of the product substrate. Notice. For example, the post-processing identification unit 130 sends a message (eg, "Please perform a repair process after the cleaning / removing process") notifying the board to be inspected of the post-processing to be executed and the execution order of the post-processing. Display on the display.

According to the configuration of this modification, it is possible for a person who inspects the product substrate to easily determine in what order the post-processing should be performed on the inspection target substrate judged to have an abnormality. it can. As a result, further efficiency of product inspection work can be expected.

<Second modification>
In the present embodiment, for example, priorities are set for each of a plurality of groups, and the post-processing specific unit 130 performs post-processing to be executed on the abnormal substrate based on the priorities set for each group. As, it may be configured to specify one post-processing. For example, when a specific post-processing includes another post-processing, the priority of the former post-processing is higher than the priority of the latter post-processing. As a specific example, a case where the substrate to be inspected is classified into a group corresponding to "repair processing" and a group corresponding to "cleaning processing" will be mentioned. Here, in repair processing such as polishing and re-plating, dirt and debris may adhere separately during the work, so be sure to perform the substrate cleaning work as a set in order to remove such dirt and debris. become. In this case, it can be said that the cleaning process is included in the repair process. Therefore, as a result of the abnormality determination unit 120 determining the abnormality based on the image of the inspection target substrate, when the inspection target substrate is classified into a group corresponding to "repair processing" and a group corresponding to "cleaning processing", at least If the repair process is executed, as a result, the cleaning process can also be executed. In such a case, for example, as shown in FIG. 7, by setting the priority of the group corresponding to the repair process higher than the priority of the group corresponding to the cleaning process, the post-processing specifying unit 130 can perform. Using the information shown in FIG. 7, it can be determined that the "repair process" having a higher priority among the "repair process" and the "cleaning process" should be executed. Information as illustrated in FIG. 7 is stored in advance in a storage area accessible to the post-processing specific unit 130, such as a memory 1030 or a storage device 1040. With such a configuration, the effect of improving the efficiency of work including post-processing can be expected.

[Third Embodiment]
The present embodiment is the same as the first embodiment described above, except for the points described below.

<Function configuration example>
FIG. 8 is a diagram illustrating the functional configuration of the substrate inspection device 10 according to the third embodiment. As shown in FIG. 8, in the present embodiment, the abnormality determination unit 120 has a plurality of classifiers 122. Each of these plurality of classifiers 122 is constructed by attribute using a plurality of learning data sets (combination of an image of an abnormality occurring on a substrate and information for identifying the type of abnormality) collected for each attribute. ing.

Here, the attributes of the substrate are, for example, the type of the substrate (for example, the type based on the substrate material, the type based on the configuration of the substrate such as the ceramic substrate or the ceramic circuit board), or the place where the substrate is manufactured (for example, the factory). Or a line in the factory, etc.). However, the attributes of the substrate are not particularly limited as long as they can affect the type of abnormality occurring on the substrate.

In the present embodiment, the abnormality determination unit 120 acquires information indicating the attributes of the substrate to be inspected (hereinafter, also referred to as “board attribute information”). Then, the abnormality determination unit 120 identifies the classifier 122 corresponding to the attribute indicated by the board attribute information as "a classifier used when determining the presence or absence of an abnormality in the substrate to be inspected". Then, the abnormality determination unit 120 determines the presence or absence of an abnormality in the inspection target substrate by using the identified classifier 122.

<Processing flow>
The flow of processing executed by the substrate inspection apparatus 10 of this embodiment will be described with reference to FIG. FIG. 9 is a flowchart showing a flow of processing executed by the substrate inspection device 10 of the third embodiment.

First, the abnormality determination unit 120 acquires the data (board attribute information) of the board to be inspected (S302). For example, the abnormality determination unit 120 receives input of information (board attribute information) indicating the type and manufacturing location of the board to be inspected via an input device (keyboard or touch panel) connected to the board inspection device 10.

The abnormality determination unit 120 selects a classifier corresponding to the attribute indicated by the input board attribute information from the plurality of classifiers 122 (S304). For example, the abnormality determination unit 120 selects a classifier corresponding to the attribute indicated by the board attribute information acquired in the process of S302 by using a table (example: FIG. 10) that stores the attribute corresponding to each classifier. Can be done. FIG. 10 is a diagram showing an example of information for associating the classifier with the attribute corresponding to the classifier. The information exemplified in FIG. 10 is stored in advance in a storage area such as a memory 1030 or a storage device 1040. For example, when the attribute indicated by the substrate attribute information is "ceramic substrate", the abnormality determination unit 120 is constructed for inspection of the ceramic substrate based on the information illustrated in FIG. Select the identified classifier A. Further, when the attribute indicated by the board attribute information is "ceramic circuit board", the abnormality determination unit 120 is a classifier constructed for inspection of the ceramic circuit board based on the information illustrated in FIG. Select B. Further, when the attribute indicated by the substrate attribute information is neither the "ceramic substrate" nor the "ceramic circuit board", the abnormality determination unit 120 is a general-purpose identification based on the information illustrated in FIG. Select vessel C.

After that, when the image acquisition unit 110 acquires the image of the inspection target substrate (S306), the abnormality determination unit 120 inputs the image of the inspection target substrate to the classifier 122 selected in the process of S304 (S308). .. Then, the abnormality determination unit 120 determines whether or not the inspection target substrate is an abnormality-existing substrate based on the output result obtained from the classifier 122 by inputting the image of the inspection target substrate ( S310). When an abnormality is detected in the inspection target substrate (S310: YES), the abnormality determination unit 120 classifies the group to be classified of the inspection target substrate (abnormality-existing substrate) determined to be abnormal into the classifier 122. Is specified based on the output result (type of abnormality) of (S312). Then, the abnormality determination unit 120 notifies a person engaged in the inspection work of the product substrate of the information about the group specified in the process of S312 via, for example, a display or a lamp connected to the substrate inspection device 10 (S314). .. On the other hand, when no abnormality is detected in the inspection target substrate (S310: NO), the abnormality determination unit 120 determines the inspection target substrate as a passing product (S316). Substrates judged to be acceptable products are shipped as products after undergoing packing work. The processing flow from S306 to S316 is the same as the processing flow from S102 to S112 in FIG.

As described above, in the present embodiment, the classifier is switched for each attribute of the substrate to be inspected, and the presence or absence of abnormality in the substrate is determined. Here, the attributes of the substrate can affect the types of anomalies that can occur on the substrate. For example, the types and tendencies of abnormalities that occur vary depending on the material and composition of the substrate. In addition, there are considerable individual differences in the manufacturing machines used at the manufacturing site, and the types and tendencies of abnormalities that occur on the substrate may differ from one manufacturing machine to another. Therefore, as described in the present embodiment, by appropriately selecting a classifier that matches the attributes of the substrate to be inspected, the effect of improving the accuracy of determining the presence or absence of abnormality in the substrate can be expected.

10 Board inspection device 110 Image acquisition unit 120 Abnormality judgment unit 122 Identifier 130 Post-processing identification unit 1010 Bus 1020 Processor 1030 Memory 1040 Storage device 1050 Input / output interface 1052 Image pickup device 1054 Input / output device 1060 Network interface

Claims (14)

An image acquisition unit that acquires multiple images generated by shooting the board under different conditions,
By processing the plurality of images using a classifier generated by machine learning,
An abnormality determination unit that determines the presence or absence of an abnormality on the substrate,
A substrate inspection device equipped with. The abnormality determination unit classifies the abnormal substrate, which is the substrate determined to have an abnormality, into a plurality of groups.
The substrate inspection apparatus according to claim 1. A post-processing identification unit is provided, which specifies the post-processing to be performed on the abnormal substrate by using the classified group.
The substrate inspection apparatus according to claim 2. Priority is set for each of the plurality of groups.
The post-processing specifying unit specifies one post-processing as the post-processing to be performed on the abnormal substrate based on the priority.
The substrate inspection apparatus according to claim 3. Priority is set for each of the plurality of groups.
The post-processing specifying unit determines the execution order of post-processing to be executed on the abnormal substrate based on the priority.
The substrate inspection apparatus according to claim 3. The abnormality determination unit
Identify the type of abnormality that the substrate with abnormality has,
The group to which the abnormal substrate belongs is determined based on the type of the abnormality.
The substrate inspection apparatus according to any one of claims 2 to 5. The abnormality determination unit
When the substrate is a ceramic substrate, cracks, presence of dirt / foreign matter, scratches, chips, or irregularities are determined as abnormalities of the substrate.
When the board is a circuit board, dirt, scratches, or irregularities on the surface are determined as abnormalities on the board.
The substrate inspection apparatus according to any one of claims 1 to 6. The post-processing to be performed on the abnormal substrate, which is the substrate determined to have an abnormality, includes any one of (a) to (d) described below.
The substrate inspection apparatus according to any one of claims 1 to 7.
(A) Use the abnormal substrate as a product substrate (b) Remove foreign matter adhering to the abnormal substrate to make a product substrate (c) Repair the abnormality of the abnormal substrate to make a product substrate (d) ) Discard the abnormal board as a rejected product. The post-treatment of (b) includes a step of cleaning and removing foreign substances with a predetermined gas or liquid.
The substrate inspection apparatus according to claim 8. The post-treatment of (c) includes a repair step by heat or polishing.
The substrate inspection apparatus according to claim 8 or 9. The classifier is constructed for each attribute using learning data according to the attribute of the substrate.
The abnormality determination unit
The board attribute information indicating the attribute of the board is acquired, and the presence or absence of an abnormality of the board is determined by using the classifier corresponding to the attribute indicated by the board attribute information.
The substrate inspection apparatus according to any one of claims 1 to 10. The substrate inspection apparatus according to any one of claims 1 to 11, wherein the attributes of the substrate include at least one of the type of the substrate and the place where the substrate is manufactured. The computer
Acquire multiple images generated by shooting the board under different conditions,
By processing the plurality of images using a classifier generated by machine learning,
Judging whether or not there is an abnormality in the substrate,
Substrate inspection method including. A program that causes a computer to execute the substrate inspection method according to claim 13.

Images