Product inspection system and product inspection method

Abstract

To provide a technique capable of determining whether a product is good with higher precision using spectral analysis.SOLUTION: A product inspection system comprises a spectrum measurement part which measures the spectrum of a product being conveyed, and a propriety determination part 44 which determines whether the product is good based upon the output obtained by inputting spectrum data on the product measured by the spectrum measurement part to a learnt model generated by machine learning. The learnt model is generated by the machine learning using teacher data including spectrum data measured in mutually different postures.SELECTED DRAWING: Figure 11

Description

[Technical field]

[0001] The present invention relates to a product inspection system and a product inspection method. [Background technology]

[0002] Patent Document 1 discloses a technology for quickly determining the quality of a product by non-destructive analysis using spectroscopic analysis. According to Patent Document 1, the spectrum of a moving product is measured and the quality of the moving product can be determined without stopping the product, thereby realizing high-throughput quality determination of the product. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2020-159971 A Summary of the Invention [Problem to be solved by the invention]

[0004] There is nothing better than being able to judge the quality of a product with higher accuracy. After extensive research, the inventors of the present invention came up with a technology that can realize more accurate quality judgment of a product using spectroscopic analysis.

[0005] The present invention has been made in consideration of the above circumstances, and has an object to provide a technique that can realize more accurate pass / fail determination of products using spectroscopic analysis. [Means for solving the problem]

[0006] In order to solve the above problems, a product inspection system according to an embodiment of the present invention includes a spectrum measurement unit that measures the spectrum of a product being transported, and a quality determination unit that determines the quality of the product based on an output obtained by inputting the spectrum data of the product measured by the spectrum measurement unit into a trained model generated by machine learning. The trained model is generated by machine learning using teacher data including spectrum data measured in different postures.

[0007] Another aspect of the present invention is a product inspection method, which includes the steps of measuring a spectrum of a product being transported, and judging whether the product is good or bad based on an output obtained by inputting the measured spectral data of the product into a trained model generated by machine learning. The trained model is generated by machine learning using training data including spectral data measured in different postures.

[0008] Any combination of the above components, and any transformation of the present invention into a method, device, system, recording medium, computer program, etc., are also effective as aspects of the present invention. Effect of the Invention

[0009] According to the present invention, it is possible to realize a more accurate quality determination of a product using spectroscopic analysis. [Brief description of the drawings]

[0010] [Figure 1] 1 is a diagram showing a configuration of a product inspection system according to an embodiment; [Diagram 2] 2 is a perspective view showing how the conveying mechanism in FIG. 1 conveys tablets. FIG. [Diagram 3] FIG. 2 is a diagram showing a state in which the jig in FIG. 1 is used. [Figure 4] 2 is a diagram illustrating the posture of a tablet in the conveying mechanism of FIG. 1. FIG. [Diagram 5] 2 is a diagram illustrating the posture of a tablet in the conveying mechanism of FIG. 1. FIG. [Figure 6] 2 is a diagram illustrating the posture of a tablet in the conveying mechanism of FIG. 1. FIG. [Figure 7] 2 is a diagram illustrating the posture of a tablet in the conveying mechanism of FIG. 1. FIG. [Figure 8] 2 is a diagram illustrating the posture of a tablet in the conveying mechanism of FIG. 1. FIG. [Figure 9] 2 is a diagram illustrating the posture of a tablet in the conveying mechanism of FIG. 1. FIG. [Figure 10] 2 is a diagram illustrating the posture of a tablet in the conveying mechanism of FIG. 1. FIG. [Figure 11] 2 is a block diagram showing a control system of the product inspection system of FIG. 1. [Figure 12] FIG. 11 is a flow diagram illustrating the operation of the preparation phase. [Figure 13] FIG. 11 is a flow diagram illustrating the operation of determining pass / fail in the inspection phase. [Figure 14] FIG. 13 is a diagram showing a configuration of a product inspection system according to a modified example. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0011] The product inspection system of the embodiment measures the spectrum of a product by spectroscopic analysis, inputs the spectrum into a trained model (calibration model) to estimate the content of a specific component in the product, and judges the quality of the product based on the estimated content of the specific component. The "trained model" is a model trained by machine learning that has been trained to estimate (output) the content of a specific component contained in a product when the spectral data of the product is input. The "specific component" is a specific component contained in a product, and is not limited to a particular component, and may be the component that has the greatest impact on the quality of the product, or the component that is contained in the product in the greatest amount.

[0012] In a product inspection system, in order to inspect a product for quality at high speed (for example, within 10 milliseconds per product), the product is irradiated with light as it moves by a conveying mechanism, and its spectrum is measured without stopping the movement of the product.

[0013] However, because the products are arranged on the conveying mechanism at high speed, the posture of the arranged products may differ from one product to another, depending on how the products are arranged.

[0014] In addition, the conveying mechanism changes over time with use. For example, if the conveying mechanism includes a conveying belt, the conveying belt stretches and sags with use. Therefore, the position of the product on the conveying mechanism may change over time.

[0015] If the attitude of the product on the conveying mechanism is different, the attitude of the product relative to the measurement unit that measures the spectrum will also be different, which will affect the measured spectrum and thus the accuracy of the trained model's estimation of the content of specific components, and ultimately the accuracy of the pass / fail judgment.

[0016] In this embodiment, a trained model is generated that takes into consideration that the orientation of the product when the spectrum is measured may differ for each product or over time. This makes it possible to estimate the content of a specific component in a product with higher accuracy, and to achieve high throughput and high-precision pass / fail judgment.

[0017] More specifically, in the "preparation phase", the product inspection system intentionally measures the spectrum of a product whose content of a specific component is known in various positions, and uses the measured spectrum as teacher (learning) data to generate a trained model. In the "inspection phase", the product inspection system measures the spectrum of a product being transported by a transport mechanism without stopping the product, and inputs the measured spectrum into the trained model generated in the preparation phase to estimate the content of a specific component in the product and determine its pass / fail.

[0018] In the following, the case where the product is a tablet will be taken as an example. The tablet is typically a drug, particularly a pharmaceutical product, but is not limited thereto, and may be, for example, a health food or a supplement.

[0019] The product may be a solid phase product other than tablets, as long as the content of a specific ingredient affects the quality. In this case, the technical concept described below can be applied to the other product to the extent that there is no contradiction, and therefore "tablets" in the description below should be read as the other product.

[0020] 1 is a diagram showing the configuration of a product inspection system 1 according to an embodiment. The product inspection system 1 includes a spectrum measurement unit 10, a transport mechanism 12, an external exclusion device 14, a jig 16, and an information processing device 18.

[0021] The spectrum measuring unit 10 irradiates light onto the tablet T and measures the spectrum of the transmitted light. In the preparation phase, the spectrum measuring unit 10 measures the spectrum of the tablet T held in the jig 16 as described below with reference to FIG. 3. This tablet T is a tablet with a known content of a specific component, and is also called a sample tablet T. In the inspection phase, the spectrum measuring unit 10 measures the spectrum of the tablets T conveyed one after another by the conveying mechanism 12 without stopping the movement as shown in FIG. 1. These tablets T are tablets to be inspected with unknown content of a specific component.

[0022] The spectrum measuring unit 10 includes an irradiating unit 20 and a light receiving unit 22. The irradiating unit 20 irradiates the tablet with measurement light L1. The light receiving unit 22 receives (detects) transmitted light L2 from the tablet T when the irradiating unit 20 irradiates the tablet T with the measurement light L1.

[0023] An example of the irradiation unit 20 and the light receiving unit 22 will be described below. It goes without saying that the irradiation unit 20 and the light receiving unit 22 are not limited to the following configuration. The irradiation unit 20 may include a broadband pulse light source and a pulse stretching element that stretches the pulse width of the emitted pulse light so that the relationship between the wavelength in one pulse and the elapsed time becomes one to one. The pulse stretching element may include an arrayed waveguide grating that spatially divides the pulse light emitted from the pulse light source according to the wavelength, and a number of fibers corresponding to the number of wavelengths divided by the arrayed waveguide grating.

[0024] Returning to FIG. 1, the measurement light L1 is irradiated onto the first surface (the bottom surface in FIG. 1) of the tablet T, passes through the tablet T, and is emitted from the second surface (top surface) as transmitted light (hereinafter also referred to as object light) L2. The spectrum of the measurement light L1 is I L1 (λ), the transmittance X(λ) of the object light L2 is wavelength-dependent, and the spectrum I L2 (λ) is expressed by the following equation (1). I L2 (λ)=X(λ)×I L1 (λ) (1)

[0025] The light receiving unit 22 is provided on the opposite side of the irradiating unit 20 across the conveying mechanism 12, and detects the diffuse transmitted light emitted from the second surface of the tablet T. The light receiving unit 22 includes a photodetector that detects the diffuse transmitted light of the tablet T as object light L2. In addition to the photodetector, the light receiving unit 22 may include an A / D converter and a focusing optical system, but these are omitted in FIG. 1. The photodetector is a photoelectric conversion element that converts an optical signal into an electrical signal, and examples of the photodetector include a photodiode, an avalanche photodiode, a phototransistor, a photomultiplier tube (photomultiplier) that utilizes the photoelectric effect, and a photoconductive element that utilizes a change in electrical resistance due to light irradiation.

[0026] The spectrum measuring unit 10 calculates the spectrum I of the object light L2 based on the output signal of the light receiving unit 22. L2 Then, the spectrum I of the measurement light L1 is generated. L1 (λ) and the spectrum of the object light L2 L2 Based on (λ), the transmittance X(λ) of tablet T is calculated using the following formula (2). X(λ)=I L2 (λ) / I L1 (λ) (2)

[0027] As a modified example, the spectrum measuring unit 10 may measure the spectrum of reflected light from the tablet T. In this case, the light receiving unit 22 may be disposed, for example, on the same side as the irradiating unit 20 with respect to the conveying surface 12a of the conveying mechanism 12, and may receive the reflected light from the tablet T when the irradiating unit 20 irradiates the tablet T with the measurement light L1.

[0028] As a further modification, the spectrum measuring unit 10 may measure the spectrum of both the transmitted light L2 and the reflected light from the tablet T.

[0029] The measurement light L1 is light capable of estimating the content of a specific component in the tablet T. The measurement light L1 is not particularly limited, but is, for example, light including a wavelength range in the near infrared region, specifically light having a wavelength range of 1000 nm or more and 1300 nm or less, more preferably light having a wavelength range of 1100 nm or more and 1200 nm or less. The measurement light L1 may be pulsed light.

[0030] The conveying mechanism 12 is a mechanism that conveys the tablets T in the direction in which the tablets T are arranged in the inspection phase. The conveying mechanism 12 conveys the tablets T from left to right in FIG. 1. The tablets T may be sucked by a suction mechanism (not shown) and adsorbed to the conveying surface 12a. In this case, the tablets are stably conveyed.

[0031] Hereinafter, the direction in which the conveying mechanism 12 conveys the tablet T will be referred to as the conveying direction x, the direction perpendicular to the conveying direction x will be referred to as the width direction y, and the direction perpendicular to both the conveying direction x and the width direction will be referred to as the height direction z.

[0032] FIG. 2 is a perspective view showing how the conveying mechanism 12 conveys the tablet T. As an embodiment, the conveying mechanism 12 includes a pair of conveying belts 52 as shown in FIG. 2. The pair of conveying belts 52 are arranged at a predetermined interval in the width direction y. The pair of conveying belts 52 are endless belts and are wound around a plurality of rollers (not shown) whose rotation axes face the width direction y. The tablet T is placed on the pair of conveying belts 52 so as to straddle them. When a driving unit (not shown), such as a motor, drives the rollers, the pair of conveying belts 52 rotates accordingly, and the tablet T is conveyed. The measurement light L1 from the irradiation unit 20 is irradiated onto the tablet T through between the pair of conveying belts 52.

[0033] Returning to Fig. 1, the information processing device 18 comprehensively controls the product inspection system 1. In the preparation phase, the information processing device 18 uses training data, which will be described in detail later, to generate a trained model used to determine the quality of the tablets T. In addition, in the inspection phase, the information processing device 18 uses the trained model to determine the quality of the tablets T, as will be described in detail later.

[0034] The system exclusion device 14 excludes only tablets T that are determined to be defective in the quality judgment during the inspection phase. The configuration of the system exclusion device 14 is not particularly limited, and may be configured using known or future available technology.

[0035] 3 is a diagram showing the state of the jig 16 in FIG. 1 when it is used. FIG. 3 is a diagram showing the spectrum measurement unit 10 and its periphery in the transport direction x. The jig 16 is provided between the irradiation unit 20 and the light receiving unit 22. In this example, the transport mechanism 12 is retreated from a position where it crosses between the irradiation unit 20 and the light receiving unit 22. Alternatively, the spectrum measurement unit 10 is retreated from a position where the transport mechanism 12 crosses between the irradiation unit 20 and the light receiving unit 22.

[0036] The jig 16 is used to hold the sample tablet T in various positions. More specifically, the jig 16 is used to reproduce the difference in position for each tablet T when measuring the spectrum, which will be described later with reference to Figs. 4 to 8. In the preparation phase, with the sample tablet T held in a desired position by the jig 16, the spectrum to be used as teacher data is measured by the spectrum measurement unit 10.

[0037] For example, the jig 16 includes a disk 16a that rotates around a rotation axis R. By rotating the disk 16a holding the tablet T, a state in which the tablet T is conveyed at a conveying speed by the conveying mechanism 12 is reproduced. In this state, the spectrum of the tablet T held on the disk 16a, which is used as teacher data, is measured by the spectrum measuring unit 10.

[0038] In addition, in the measurement of the spectrum in the preparation phase, i.e., the measurement of the spectrum to be used as teacher data, the spectrum may be measured by a spectrum measurement unit (not shown) other than the spectrum measurement unit 10 that has the same configuration as the spectrum measurement unit 10. The other spectrum measurement unit may be provided at a position where the transport mechanism 12 does not cross between its irradiation unit and light receiving unit.

[0039] Next, different postures for each tablet T in the conveying mechanism 12 will be described with reference to Figs. 4 to 8. Note that "different postures" here also include cases where the positions in the width direction y and height direction z are different. Also, here, a case where the conveying mechanism 12 is configured to include a pair of conveying belts 52 will be described.

[0040] 4 is a view of the conveying mechanism 12 viewed in a direction perpendicular to the conveying surface 12a. In this example, the positions of the tablets T in the width direction y are different from one another. Although it depends on how the tablets T are arranged, because the tablets T are arranged on the conveying mechanism 12 at high speed (for example, 10 milliseconds or less per tablet), a deviation in position in the width direction may occur.

[0041] 5(a) to (c) are views of the conveying mechanism 12 viewed in the width direction y. In FIG. 5(a) to (c), at the irradiation position where the measurement light L1 is irradiated onto the tablet T, the positions of the tablet T in the height direction z relative to a predetermined reference conveying surface S are different from each other. The reference conveying surface S is, for example, a surface that matches the conveying surface before the conveying mechanism 12 is deteriorated over time, although it is not particularly limited thereto. In FIG. 5(a), the conveying surface matches the reference conveying surface S. In FIG. 5(b) and (c), for example, the conveying belt stretches and bends due to deterioration over time, so that the conveying surface 12a is lower than the reference conveying surface S. In FIG. 5(a) to (c), the heights of the conveying surface 12a at the irradiation positions are different from each other, so that the heights relative to the reference conveying surface S are different from each other.

[0042] 6(a) and (b) are views of the conveying mechanism 12 viewed in the conveying direction x. In FIG. 6(a) and (b), at the irradiation position where the measurement light L1 is irradiated onto the tablet T, the angle θy of the tablet T with respect to a predetermined conveying reference plane S, which is the angle θy around an axis parallel to the conveying direction x, is different from one another. The reference conveying plane S is not particularly limited, but is, for example, a plane that matches the conveying surface before the conveying mechanism 12 is subject to aging. For example, when the curved outer surface portion of the tablet T comes into contact with the conveying surface 12a, the angle θy of the tablet T with respect to the reference conveying plane S may differ from one tablet T to another.

[0043] 7(a) and (b) are views of the conveying mechanism 12 as viewed in the conveying direction x. In Fig. 7(a) and (b), similarly to Fig. 6(a) and (b), the angle θy of the tablet T at the irradiation position is different from each other. For example, the angle θy may differ depending on the difference in the progress of aging of the pair of conveying belts 52.

[0044] 8(a) and (b) are views of the conveying mechanism 12 viewed in the width direction y. In FIG. 8(a) and (b), at the irradiation position where the measurement light L1 is irradiated onto the tablet T, the angle θx of the tablet T with respect to a predetermined conveying reference plane S, which is the angle θx around an axis parallel to the width direction y, is different from one another. The reference conveying plane S is not particularly limited, but is, for example, a plane that matches the conveying surface before the conveying mechanism 12 is subject to aging. For example, when the curved outer surface portion of the tablet T comes into contact with the conveying surface 12a, the angle θx of the tablet T with respect to the reference conveying plane S may differ from one tablet T to another.

[0045] 9(a) and (b) are views of the conveying mechanism 12 viewed in the width direction y. In Fig. 9(a) and (b), similarly to Fig. 8(a) and (b), the angle θx of the tablet T at the irradiation position is different from each other. For example, the angle θx may differ due to aging of the conveying belt 52.

[0046] Fig. 10 is a view of the conveying mechanism 12 viewed in a direction perpendicular to the conveying surface 12a. In Fig. 8, the tablets T have a rotationally asymmetric shape, and the orientations around an axis perpendicular to the conveying surface 12a (in other words, around an axis parallel to the height direction z) are different from each other. Here, the tablets T have a dividing line L, and when viewed in a direction perpendicular to the conveying surface 12a, the extending directions of the dividing line L are different from each other.

[0047] Although the case where the transport mechanism 12 includes the pair of transport belts 52 has been described here, even if the transport mechanism 12 has another configuration, at least one of the postures shown in FIGS. 4 to 10 may differ from that shown in FIG.

[0048] Fig. 11 is a block diagram showing the control system of the product inspection system 1. Each block shown here can be realized in hardware terms by elements and mechanical devices such as a computer's CPU (central processing unit), and in software terms by a computer program, etc., but here, functional blocks realized by the cooperation of these are depicted. Therefore, those skilled in the art who have read this specification will understand that these functional blocks can be realized in various ways by combining hardware and software.

[0049] The information processing device 18 includes a spectrum data receiving unit 40, a model generating unit 42, a quality determining unit 44, and a model storage unit 46. Note that, as for the components of the information processing device 18, only the components of interest in this embodiment are shown.

[0050] There is no limit to the number of physical devices (housings) constituting the information processing device 18, and the information processing device 18 may be realized as a single device or may be realized by cooperation of multiple devices. For example, the information processing device 18 may be realized by cooperation of a first device and a second device, the first device may include a spectrum data receiving unit 40, a model generating unit 42, and a model storage unit 46, and the second device may include a pass / fail determining unit 44.

[0051] The model storage unit 46 stores a trained model (calibration model) generated by the model generation unit 42 described later.

[0052] In the preliminary phase, the spectrum data receiving unit 40 receives the spectrum data of the sample tablet T measured by the spectrum measuring unit 10. Note that the spectrum of the sample tablet T may be measured by a spectrum measuring unit other than the spectrum measuring unit 10 as described above, and in this case, the spectrum data receiving unit 40 receives the spectrum data of the sample tablet T measured by the other spectrum measuring unit.

[0053] Furthermore, the spectrum data receiving section 40 receives the spectrum data of the tablets T sequentially measured by the spectrum measuring section 10 in the inspection phase.

[0054] In the preparation phase, the model generation unit 42 performs machine learning using training data to generate a trained model that inputs the spectral data of the tablet T and outputs (estimates) the content of a specific component in the tablet T. The trained model may be configured using any known or future available machine learning method. The model generation unit 42 stores the generated trained model in the model storage unit 46. The training data will be described later.

[0055] In the inspection phase, the quality determination unit 44 uses the trained model stored in the model storage unit 46 to determine the quality of the tablet T. In detail, the quality determination unit 44 inputs the spectral data received by the spectral data receiving unit 40 to the trained model, thereby causing the trained model to estimate (output) the content of the specific component in the tablet T.

[0056] Next, the quality determination unit 44 determines whether the tablet T is good or bad based on the content of the specific component estimated by the trained model. If the ratio (hereinafter referred to as the specified content) of the estimated content of the characteristic component (i.e., the content of the specific component actually contained in the tablet T) to the content of the specific component to be contained in the tablet T is within a predetermined threshold range, for example, if the specified content is 98% or more and 102% or less, the quality determination unit 44 determines the tablet T as a good product, and if the specified content is outside the threshold range, the quality determination unit 44 determines the tablet T as a defective product. The quality determination unit 44 may determine the quality of the tablet T by further considering at least one other element in addition to the content of the specific component. The other element may be, for example, the appearance of the tablet T. In this case, the quality determination unit 44 determines the tablet T as a defective product if the appearance of the tablet T is defective even if the specified content is within a predetermined threshold range. Whether or not the appearance of tablet T is defective may be determined by performing image analysis on an image of tablet T photographed by a photographing unit not shown in the figure. For example, the appearance of tablet T may be determined to be defective if tablet T has deformation such as chips or cracks.

[0057] The outside-system exclusion device 14 includes an exclusion control unit 48. The exclusion control unit 48 performs overall control of the outside-system exclusion device 14. The exclusion control unit 48 controls the outside-system exclusion device 14 to exclude tablets T determined to be defective by the quality determination unit 44 from the system. Since tablets T determined to be defective are automatically excluded, accidents in which defective products are shipped can be prevented.

[0058] Next, the "teaching data" will be described. The teaching data is a plurality of data sets for a plurality of sample tablets T whose contents of a specific component are known, each of which includes the content of the specific component and spectral data.

[0059] The multiple sample tablets T include sample tablets T with various contents, i.e., sample tablets T with different contents of a specific component. Preferably, the multiple sample tablets T include a sample tablet T with a specific component content of 100%.

[0060] Preferably, the multiple sample tablets T include sample tablets T that can cover the range of the expected content of the specific component of the tablet T. That is, preferably, the multiple sample tablets T include sample tablets T whose content of the specific component is within the expected range, sample tablets T whose content of the specific component is lower than the expected range, and sample tablets T whose content of the specific component is higher than the expected range.

[0061] Variations in the content of the specific component contained in the multiple sample tablets T may affect the accuracy of the generated model. Therefore, for example, the variation in the content of the specific component contained in the multiple sample tablets T may be determined by trial and error. For example, when the expected content range of the specific component of the tablet T is 90% to 110%, the multiple sample tablets T may include sample tablets T with a content in 2% increments in the range of 88% to 112%, that is, sample tablets T with a content of the specific component of 88%, 90%, 92%, ..., 112%.

[0062] The multiple data sets include data sets of spectrum data measured in different orientations. The different orientations are orientations that are different from each other in at least one of the following: the position in the width direction y, the position in the height direction z, the angle θy of the tablet T with respect to a predetermined transport reference plane (i.e., the angle θy around an axis parallel to the transport direction x), the angle θx of the tablet T with respect to a predetermined transport reference plane (i.e., the angle θx around an axis parallel to the width direction y), and the orientation around an axis perpendicular to the transport surface 12a (i.e., around an axis parallel to the height direction z), as described in Figs. 4 to 10.

[0063] The multiple data sets may include only one data set measured in one orientation for each sample tablet T, or may include multiple data sets measured in multiple orientations for each sample tablet T.

[0064] The above is the basic configuration of the product inspection system 1. Next, the operation will be described.

[0065] 12 is a flow diagram explaining the operation of the preparation phase. First, the spectra of multiple sample tablets T whose contents of specific components are known are measured in various postures (S10). Next, a trained model to be used for determining whether the tablets T are good or bad is generated using training data including the measured spectral data of the sample tablets T (S12).

[0066] Fig. 13 is a flow diagram explaining the operation of pass / fail judgment in the inspection phase. The process of Fig. 13 is repeatedly performed at a predetermined cycle (for example, a predetermined cycle of 10 milliseconds or less), in other words, for each tablet that is transported one after another. First, the spectrum of the tablet T transported and moving by the transport mechanism 12 is measured without stopping the movement (S20). Next, the measured spectrum data is input to the trained model to estimate the content of a specific component in the tablet T (S22). Next, the quality of the tablet T is judged based on the estimated content of the specific component (S24).

[0067] According to this embodiment, in the preparation phase, a trained model is created using teacher data including spectral data for a plurality of sample tablets T whose contents of specific components are known and which are intentionally measured in various postures, and in the inspection phase, the trained model is used to estimate the contents of specific components in the tablets and to judge the quality of the tablets. In other words, since the contents of specific components are estimated based on a trained model that takes into consideration in advance that the posture of the tablet T when measuring the spectrum may differ for each tablet T or over time, the contents of specific components can be estimated with higher accuracy, and the quality of the tablet T can be judged with higher accuracy.

[0068] The present invention has been described above based on the embodiment. This embodiment is merely an example, and it will be understood by those skilled in the art that various modifications are possible in the combination of each component and each treatment process, and that such modifications are also within the scope of the present invention.

[0069] (Variation 1) In the embodiment, the jig 16 is used to hold the sample tablet T in various postures, and the spectrum of the sample tablet T in the various postures is measured to be used as teacher data. However, it is sufficient to be able to measure the spectrum of the sample tablet T in various postures, and the present invention is not limited to measuring the spectrum using the jig 16.

[0070] 14 is a diagram showing the configuration of a product inspection system 101 according to a modified example. The following mainly describes the differences from the embodiment.

[0071] The product inspection system 101 includes a spectrum measuring unit 10, a transport mechanism 12, an outside exclusion device 14, an information processing device 18, and a posture detection unit 108. That is, the product inspection system 101 of this modified example does not include the jig 16.

[0072] In the preparation phase of this modified example, similarly to the inspection phase, the transport mechanism 12 transports the sample tablets T one after another, and the spectrum measurement unit 10 measures the spectra of the sample tablets T transported one after another by the transport mechanism 12 without stopping their movement.

[0073] In the preparation phase, the posture detection unit 108 detects the posture of the sample tablet T being transported by the transport mechanism 12. The posture detection unit 108 may detect the posture of the sample tablet T upstream of the spectrum measurement unit 10 as shown in the figure, may detect the posture of the sample tablet T downstream of the spectrum measurement unit 10, or may detect the posture of the sample tablet T at a position where the spectrum is measured by the spectrum measurement unit 10 (for example, an irradiation position where the measurement light L1 from the irradiation unit 20 is irradiated onto the sample tablet T).

[0074] The configuration of the attitude detection unit 108 is not particularly limited. The attitude detection unit 108 may be configured to include at least one camera, for example. In this case, the attitude detection unit 108 may be capable of photographing the tablet T from three directions, the conveying direction x, the width direction y, and the height direction z. In addition, for example, the attitude detection unit 108 may be configured to include an ultrasonic sensor. In any case, the attitude detection unit 108 may be configured to detect the attitude of the sample tablet T.

[0075] The multiple data sets of teacher data in this modified example include a data set of the spectral data of the sample tablet T measured while being transported by the transport mechanism 12 as described above. The multiple data sets of teacher data are not particularly limited, and may be selected by the user from the spectral data of the measured sample tablet T. The user may select multiple data sets of teacher data based on the detection result of the posture of the sample tablet T by the posture detection unit 108, for example, by referring to an image of the sample tablet T captured by at least one camera as the posture detection unit 108, so as to include a data set of spectral data measured in various postures. For example, the user may select multiple data sets of teacher data so as to include a data set of spectral data measured in postures in which at least one of the following is different from each other: the position in the width direction y, the angle θy of the tablet T with respect to a predetermined transport reference plane, the angle θx of the tablet T with respect to a predetermined transport reference plane, and the orientation around an axis perpendicular to the transport surface 12a.

[0076] According to this modification, it is possible to achieve the same effects as those of the embodiment.

[0077] (Variation 2) In the embodiment, the spectrum measuring section 10 and the information processing device 18 are communicatively connected to each other. However, the spectrum measuring section 10 may have the information processing device 18 provided therein.

[0078] (Variation 3) In the embodiment, a case has been described in which the trained model used in pass / fail judgment is trained to output (estimate) the content of a specific component in a tablet when spectral data of the tablet is input, but this is not limited to this, and the trained model may also be trained to output (estimate) the pass / fail of the tablet in terms of the content of the specific component. [Explanation of symbols]

[0079] 1 product inspection system, 10 spectrum measurement unit, 44 pass / fail judgment unit.

Claims

1. a spectrum measuring unit for measuring a spectrum of a product being conveyed; a quality determination unit that determines the quality of the product based on an output obtained by inputting the spectrum data of the product measured by the spectrum measurement unit into a trained model generated by machine learning; and Equipped with The product inspection system, in which the trained model is generated by machine learning using training data including spectral data measured in different postures.

2. The product inspection system of claim 1 , wherein the trained model is configured to output a content of a specific component.

3. The product inspection system according to claim 1 , wherein the spectral data measured in different postures include spectral data measured at different positions in a direction perpendicular to the conveying direction.

4. The product inspection system according to claim 1 , wherein the spectral data measured in different attitudes includes spectral data measured in attitudes having different heights relative to a reference conveying surface.

5. The product inspection system according to claim 1 , wherein the spectral data measured at different attitudes includes spectral data measured at attitudes having different angles with respect to a reference conveying plane when viewed in the conveying direction.

6. The product inspection system according to claim 1 , wherein the spectral data measured at different attitudes includes spectral data measured at attitudes having different angles with respect to a reference conveying plane when viewed in a direction perpendicular to the conveying direction.

7. The product inspection system according to claim 1 , wherein the spectral data measured at different orientations includes spectral data measured at different orientations around an axis perpendicular to the conveying surface when the product has a rotationally asymmetric shape.

8. 2. The product inspection system according to claim 1, wherein the spectral data measured in different postures includes, when the product is a tablet having a score line, spectral data measured in postures in which the extension directions of the score line are different from each other when viewed in a direction perpendicular to the conveying surface.

9. The product inspection system according to claim 1 , wherein the quality determining unit further determines the quality of the product based on an appearance of the product.

10. measuring a spectrum of the product being delivered; A step of judging whether the product is good or bad based on an output obtained by inputting the measured spectral data of the product into a trained model generated by machine learning; Equipped with A product inspection method in which the trained model is generated by machine learning using training data including spectral data measured in different postures.

Images