Inspection monitoring device

Abstract

To provide an inspection monitoring device capable of detecting suspicious inspection conditions.SOLUTION: An inspection monitoring device comprises an evaluation unit that evaluates the suspiciousness of an inspection based on inspection data over a plurality of inspection items, and a notification unit that reports that the inspection is in a suspicious state based on the evaluation results of the evaluation unit.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to an inspection monitor. [Background technology]

[0002] The following Patent Document 1 discloses a medical image diagnostic device capable of classifying at least two types of materials in multi-energy image data simply and with high accuracy. This medical image diagnostic device includes a change unit that adaptively changes a threshold value based on multi-energy intensity information associated with pixels or voxels of the multi-energy image data, and a classification unit that classifies multiple pixels or multiple voxels of the multi-energy image data into pixels or voxels corresponding to at least two types of materials based on the changed threshold value. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2016-87469 A Summary of the Invention [Problem to be solved by the invention]

[0004] Incidentally, the above-mentioned background art is for classifying two types of substances simply and accurately in the medical field, but in the product manufacturing field, products are often classified into good and bad products in product inspection. There are cases where a product that is judged to be good in this product inspection develops a defect during actual use. Therefore, in the product manufacturing field, it is an extremely important technical issue to detect suspicious states in product inspection.

[0005] The present invention has been made in consideration of the above-mentioned circumstances, and has an object to provide an inspection monitoring device capable of detecting suspicious states in product inspection. [Means for solving the problem]

[0006] In order to achieve the above-mentioned object, the present invention adopts, as a first solution relating to an inspection monitoring device, a means of providing an evaluation unit that evaluates the suspiciousness of a test based on inspection data covering a plurality of inspection items, and an alarm unit that notifies when the test becomes suspicious based on the evaluation result of the evaluation unit.

[0007] In the present invention, as a second solution related to the test monitoring device, the above-mentioned first solution is adopted, in which the evaluation unit in the above-mentioned first solution includes an index value generation unit that generates a predetermined evaluation index for each test item based on the statistics, abnormality degree, and Jensen-Shannon divergence of test data across a plurality of test items in a predetermined evaluation period, an estimate value generation unit that generates an estimate indicating the suspiciousness of the test based on the evaluation index for the previous and following evaluation periods, and a judgment unit that judges whether to output an alarm by comparing the estimate value with an alarm threshold value, and generates an alarm signal based on the result of the judgment.

[0008] In the present invention, as a third solution related to the inspection monitoring device, a means is adopted in which, in the above-mentioned second solution, the index value generation unit calculates the evaluation index by summing the variation amount of the statistics, the variation amount of the abnormality degree, and the variation amount of the Jensen-Shannon information amount using a predetermined weighting coefficient.

[0009] In the present invention, as a fourth solution related to the inspection monitoring device, a means is adopted in which, in the second or third solution means above, the estimated value generation unit generates as the estimated value the difference between the evaluation periods before and after the total value of the evaluation indexes for all of the inspection items.

[0010] The present invention employs, as a fifth solving means related to the inspection monitoring device, any one of the second to fourth solving means, further comprising a threshold setting unit that adaptively sets the alarm threshold value.

[0011] The present invention employs a sixth solution relating to an inspection monitoring device, in which any of the first to fifth solutions described above further includes a switching instruction unit that instructs switching from a suspect inspection device to a normal inspection device. Effect of the Invention

[0012] According to the present invention, it is possible to provide an examination monitoring device capable of detecting suspicious conditions of an examination. [Brief description of the drawings]

[0013] [Figure 1] 1 is a block diagram showing a functional configuration of an inspection monitoring device according to an embodiment of the present invention; [Diagram 2] 4 is a flowchart showing the operation of the inspection monitoring device according to the embodiment of the present invention. [Diagram 3] FIG. 2 is a schematic diagram illustrating setting of an alarm threshold in an inspection monitor according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0014] Hereinafter, an embodiment of the present invention will be described with reference to the drawings. 1, the inspection monitoring device A according to this embodiment at least includes an inspection data acquisition unit 1, a storage unit 2, an operation unit 3, a calculation unit 4, and an output unit 5. This inspection monitoring device A monitors product inspections that are sequentially performed during the manufacture of products by, for example, installing a predetermined dedicated application program (inspection monitoring program) on a general-purpose computer.

[0015] The inspection data acquisition unit 1 is a device that acquires inspection data from a product inspection device that is installed as part of the manufacturing equipment. This inspection data acquisition unit 1 sequentially acquires multiple pieces of inspection data related to multiple products that are acquired by the product inspection device every moment, based on an acquisition command input from the calculation unit 4, and stores the multiple pieces of inspection data in the storage unit 2 sequentially.

[0016] Here, the inspection data includes inspection results across multiple inspection items along with product identification data for identifying the product being inspected. The inspection data also includes at least a time stamp (inspection time data) indicating the date and time of the product inspection. In other words, the inspection data for one product includes at least the product identification data, inspection time data, and multiple inspection results for multiple inspection items.

[0017] The storage unit 2 sequentially stores such test data in the order of acquisition by the test data acquisition unit 1. The storage unit 2 also stores in advance the above-mentioned test monitoring program and test monitoring data required for executing the test monitoring program. The test monitoring program and the test monitoring data will be described in detail in the operation description below.

[0018] The operation unit 3 is a device that an operator of the manufacturing equipment scans, and is, for example, a pointing device such as a keyboard or a mouse. The operation unit 3 receives operation instructions, such as an instruction to start the inspection monitoring device A, from the operator and outputs them to the calculation unit 4. The operation unit 3 also receives instructions from the operator regarding the execution of the inspection monitoring program in the calculation unit 4 and outputs them to the calculation unit 4.

[0019] The calculation unit 4 executes the inspection monitoring program to evaluate (monitor) the product inspection based on the inspection data. That is, the calculation unit 4 reads the inspection monitoring program and the inspection monitoring data from the storage unit 2, and performs a predetermined calculation process on the inspection data based on the inspection monitoring program and the inspection monitoring data, thereby monitoring the state of the product inspection by the product inspection device.

[0020] The details of the above calculation process will be described later, but the calculation unit 4 is an evaluation unit that evaluates the suspiciousness of the test based on the test data across multiple test items. That is, the calculation unit 4 generates a predetermined evaluation index for each test item based on the statistics, abnormality degree, and Jensen-Shannon divergence of the test data across multiple test items in a predetermined evaluation period T, generates an estimated value indicating the suspiciousness of the test based on the evaluation index in the previous and next evaluation periods, adaptively sets an alarm threshold value based on the estimated value, determines whether to output an alarm based on a comparison between the estimated value and the alarm threshold value, and generates an alarm signal based on the result of this determination and outputs it to the output unit 5. Such a calculation unit 4 corresponds to the evaluation unit, index value generation unit, estimated value generation unit, threshold value setting unit, and determination unit in the present invention.

[0021] The calculation unit 4 also functions as a switching instruction unit. That is, the calculation unit 4 instructs the switching of a suspect inspection device to a normal inspection device (backup inspection device) based on the evaluation result of the product inspection based on the inspection data. The calculation unit 4 generates an instruction signal (switching signal) to instruct the switching of such an inspection device and outputs it to the output unit 5.

[0022] The output unit 5 notifies that the inspection by the inspection device has become suspicious based on the evaluation result of the calculation unit 4 (evaluation unit), i.e., the alarm signal input from the calculation unit 4. In other words, this output unit 5 is a device for notifying an operator of the manufacturing equipment of the evaluation result (monitoring result) of the product inspection in the calculation unit 4. Such an output unit 5 corresponds to the notification unit in the present invention. Furthermore, this output unit 5 instructs the operator or the inspection device to switch the inspection device based on the switching signal input from the calculation unit 4.

[0023] Next, the operation of the inspection monitoring device A according to this embodiment will be described in detail with reference to the flowchart shown in Fig. 2. Note that the series of processes according to this flowchart are arithmetic processes executed by the arithmetic unit 4 based on the inspection monitoring program.

[0024] When the inspection monitoring device A is started, the calculation unit 4 causes the inspection data acquisition unit 1 to sequentially acquire the inspection data of the product inspection device (step S1). That is, the calculation unit 4 starts the execution of the inspection monitoring program in accordance with a start instruction input by an operator manually operating the operation unit 3, and first outputs an instruction to the inspection data acquisition unit 1 to acquire the inspection data.

[0025] The inspection data acquisition unit 1 sequentially acquires the inspection data from the product inspection device based on the acquisition command. As a result, the storage unit 2 sequentially stores a plurality of pieces of inspection data relating to a plurality of products in the order in which the product inspection device performed the product inspection.

[0026] The calculation unit 4 performs a primary judgment based on the primary judgment criteria specified by the inspection monitoring data for the inspection data thus stored in the storage unit 2 (step S2). This primary judgment is to separate the products into good products (normal products) and defective products (abnormal products) by evaluating the inspection data based on the primary judgment criteria.

[0027] When this primary judgment is completed, the calculation unit 4 adaptively sets an alarm threshold value R for each evaluation period (step S3), as shown in Fig. 3. This alarm threshold value R is a judgment threshold value used to judge the suspiciousness of the product inspection (step S6), and is set for each evaluation period specified by the inspection monitoring data.

[0028] The calculation unit 4 groups the multiple test data stored in the storage unit 2 for each evaluation period. That is, the multiple test data stored sequentially in the storage unit 2 are virtually allocated to multiple evaluation periods arranged in chronological order. The multiple test data included in one evaluation period are relatively close in examination time.

[0029] The inspection monitoring data also includes the alarm threshold value (day-before alarm threshold value) used to determine the suspiciousness of the product inspection on the previous day. The calculation unit 4 sets the largest (maximum value) of the day-before alarm threshold values ​​acquired from the inspection monitoring data as the initial value of the alarm threshold value R for the current day.

[0030] The calculation unit 4 adaptively updates the warning threshold value R within a preset allowable range in the inspection monitoring data. The allowable range has an upper limit value that is half the primary judgment criterion and a lower limit value that is ¼ of the primary judgment criterion. The warning threshold value R is updated within an update allowable value specified in the inspection monitoring data.

[0031] This update tolerance defines the upper limit of the difference between the warning threshold value R before the update and the warning threshold value R after the update, and is intended to suppress excessive changes in the warning threshold value R. The calculation unit 4 updates the warning threshold value R for the next evaluation period based on the distribution of the test data for the first evaluation period on the day, within the limits of such set tolerance range and update tolerance value.

[0032] Then, the calculation unit 4 updates the warning threshold value R for the next evaluation period based on the distribution of the test data in the next evaluation period. That is, the warning threshold value R for a certain evaluation period is adaptively updated based on the distribution of the test data in the previous (past) evaluation period. Note that the process of step S3 is not limited to being performed immediately before step S4, so long as it is a step prior to the determination process of step S6.

[0033] Then, the calculation unit 4 calculates an evaluation period T n When the setting of the alarm threshold value R in (n: parameter specifying the evaluation period T) is completed, the evaluation index Y required for evaluating the product inspection for each inspection item for the evaluation period T including the present is n is calculated (step S4).

[0034] Then, the calculation unit 4 calculates an evaluation period T nBased on the test data, the calculation unit 4 calculates the statistics S, the degree of abnormality ABN, and the Jensen-Shannon divergence Djs for each test item. After calculating the statistics S, the degree of abnormality ABN, and the Jensen-Shannon divergence Djs for each test item, the calculation unit 4 temporarily stores these statistics S, the degree of abnormality ABN, and the Jensen-Shannon divergence Djs in the storage unit 2.

[0035] The statistic S is the mean value μ and standard deviation σ for multiple test data. As is well known, the degree of abnormality ABN is a value obtained by dividing the square of the difference between the test data value and the mean value μ by the square of the standard deviation σ. Furthermore, the Jensen-Shannon divergence Djs is a value obtained by the well-known formulas (1) to (3) below, and is specifically calculated by the test data value and the mean value μ.

[0036]

number

[0037] Then, the calculation unit 4 calculates the statistical quantity S, the degree of abnormality ABN, and the Jensen-Shannon divergence Djs for each test item, and then calculates the average value of the statistical quantity S (statistical mean value), the average value of the degree of abnormality ABN (abnormal mean value), and the average value of the Jensen-Shannon divergence Djs (JS mean value) for each test item.

[0038] Furthermore, by using the statistical quantity S, the degree of abnormality ABN, the Jensen-Shannon information quantity Djs, as well as the statistical mean value, the abnormal mean value, and the JS mean value, the calculation unit 4 calculates the amount of mutation of the statistical quantity S relative to the statistical mean value (statistical mutation amount ΔS), the amount of mutation of the degree of abnormality ABN relative to the abnormal mean value (abnormal mutation amount ΔABN), and the change in the Jensen-Shannon information quantity Djs relative to the JS mean value (JS mutation amount ΔDjs) for each test item.

[0039] Then, the calculation unit 4 calculates an evaluation index Y for each test item by summing up the standard deviation deviation Δσ, the abnormal mutation amount ΔABN, and the JS mutation amount ΔDjs in the statistical mutation amount ΔS using predetermined weighting coefficients W1 to W3. That is, the calculation unit 4 calculates the evaluation index Y for each test item based on the following formula (4). Y=W1·Δσ+W2·ΔABN+W3·ΔDjs (4)

[0040] The calculation unit 4 calculates an evaluation period T n Evaluation index Y for each test item n Calculating the current evaluation period T n An estimate of the suspiciousness of product testing in n is calculated (step S5).

[0041] That is, the calculation unit 4 calculates an evaluation period T n Evaluation index Y for all test items in n The sum of ΣY n Calculate the total value ΣY n The evaluation period immediately before, that is, the previous evaluation period T n-1 The estimated value E n-1 The difference between these is multiplied by the coefficient G to get the estimated value E n It is calculated as follows. E n =E n-1 +G×(ΣY n -E n-1 ) (5)

[0042] In this way, the evaluation period T n The estimated value of E n Then, the calculation unit 4 calculates the estimated value E n If the determination in step S6 is "Yes," that is, if the estimated value E n exceeds the warning threshold value R, the warning signal is output to the output unit 5. As a result, the output unit 5 issues a warning based on the warning signal (step S7).

[0043] In this way, if the determination in step S6 is "Yes", the product inspection in the evaluation period Tn including the current one is n-1 This indicates that the product inspection has changed significantly from the product inspection at the evaluation period Tn. In other words, if the determination in step S6 is "Yes", this indicates that the product inspection in the evaluation period Tn, including the present, has reached a significantly suspicious state. The warning from the output unit 5 allows the worker involved in the product inspection to recognize that the current product inspection has reached a suspicious state.

[0044] According to the test monitoring device A of this embodiment, a calculation unit 4 (index value generation unit) generates a predetermined evaluation index Y for each test item based on the statistics S of test data across multiple test items during a predetermined evaluation period T, the abnormality degree ABN, and the Jensen-Shannon divergence Djs, and n , T n-1 An estimate of the suspiciousness of a test based on the evaluation index in n A calculation unit 4 (estimated value generating unit) that generates an estimated value E n A calculation unit 4 (threshold setting unit) that adaptively sets an alarm threshold R based on the estimated value E n The device is equipped with a calculation unit 4 (judgment unit) that judges whether to output an alarm based on a comparison between the detected value and an alarm threshold R and generates an alarm signal based on the result of the judgment, and an output unit 5 (alarm unit) that issues an alarm based on the alarm signal, making it possible to detect suspicious conditions in product inspection.

[0045] If the determination in step S6 is "No", that is, if the evaluation period T n Product inspection in the immediately preceding evaluation period T n-1 If there has been no significant change since the product inspection in the evaluation period T n Since there is no suspicious situation in the product inspection at, the output unit 5 does not issue an alarm.

[0046] If the determination in step S2 is "No" and the determination in step S6 is "Yes", a switching signal is output to indicate switching from the current inspection device to the backup inspection device (step S8). n The degree of influence on each test item is judged, and a switching instruction is generated to switch the test device for the test item judged to have a relatively large influence, i.e., the suspect test device, to a spare test device (normal test device), and output to the output unit 5.

[0047] An operator or inspection device involved in the product inspection switches the inspection device currently being used for the product inspection to the spare inspection device based on the switching instruction from the output unit 5. That is, according to the processing of step S8, it is possible to quickly find and replace a suspicious inspection device, and therefore it is possible to quickly correct any inspection items for which the inspection is suspect. [Explanation of symbols]

[0048] A Inspection monitoring device 1. Inspection data acquisition section 2 Storage section 3 Control section 4 Calculation unit (evaluation unit, index value generation unit, estimate value generation unit, threshold value setting unit, judgment unit, switching instruction unit) 5 Output section (notification section)

Claims

1. an evaluation unit that evaluates the suspiciousness of a test based on test data across a plurality of test items; a notification unit that notifies the user that the test has become suspicious based on the evaluation result of the evaluation unit; An inspection monitoring device comprising:

2. The evaluation unit is an index value generating unit that generates a predetermined evaluation index for each of a plurality of test items based on statistics, an abnormality degree, and a Jensen-Shannon divergence of test data over a predetermined evaluation period; an estimate generating unit that generates an estimate indicating the suspiciousness of an examination based on the evaluation indexes in the preceding and succeeding evaluation periods; a determination unit that determines whether to output a warning by comparing the estimated value with a warning threshold value and generates a warning signal based on the result of the determination; 2. The inspection monitor of claim 1, further comprising:

3. The inspection monitoring device according to claim 2, characterized in that the index value generation unit calculates the evaluation index by summing up the amount of variation in statistics, the amount of variation in abnormality, and the amount of variation in Jensen-Shannon information using a predetermined weighting coefficient.

4. 4. The inspection monitoring device according to claim 2, wherein the estimated value generating unit generates, as the estimated value, a difference between the total value of the evaluation indexes for all of the inspection items during the evaluation period before and after the total value.

5. 4. The inspection monitor according to claim 2, further comprising a threshold setting unit that adaptively sets the warning threshold value.

6. 4. The inspection monitoring device according to claim 2, further comprising a switching instruction unit that instructs switching of the suspect inspection device to a normal inspection device.

Images