JP4825482B2 - Failure occurrence prediction device and failure occurrence prediction method - Google Patents

Description

  The present invention relates to a failure occurrence prediction device and a failure occurrence prediction method for predicting a failure occurrence time when a product is produced for each lot.

  Specifically, the present invention obtains the average value and standard deviation σ of the measurement data of a predetermined lot, and sequentially predicts the 6σ data range of the first to predetermined number lots following the predetermined lot, and uses the pth When the data range of a lot is out of the specified standard width, the occurrence time of the failure is displayed by using the value of p, so that the occurrence time of the failure can be notified to the user and the occurrence of the failure can be prevented in advance. This relates to a malfunction occurrence predicting device or the like.

  In a factory or the like, when producing a product, an error may be detected in the process of inspecting the physical characteristics and electrical characteristics of the product. Such errors often cause problems such as product out of specification. When such a malfunction occurs, if the cause is not discovered at an early stage and no countermeasure is taken, the production of the factory may be stopped and a great loss may occur.

  Conventionally, product errors are measured sequentially for each lot, and the measurement data is continuously arranged to display a graph showing changes over time, and a histogram is obtained based on the error, and the graph and the histogram are displayed side by side at the same time. Therefore, a display method of measurement data that can easily grasp the content of the error has been proposed (see Patent Document 1).

JP-A-8-5410

  However, although the measurement data display method described above can grasp the variation in the lot where the defect occurred and the error that causes the defect when the defect occurs, it predicts the occurrence of the defect and Nothing was suggested about preventing it.

  An object of the present invention is to inform the user of the time of occurrence of a defect before the occurrence of the defect in the production process for each lot of products so that the occurrence of the defect can be prevented.

The concept of this invention is
A measurement data input unit for inputting measurement data obtained by measuring physical characteristics or electrical characteristics of a product for each lot;
Based on the measurement data input in the measurement data input unit, a calculation unit that calculates the average value and standard deviation σ of the measurement data for each lot;
The 6σ data range from the first lot to the predetermined number of lots following the predetermined lot for which the average value and standard deviation σ of the measurement data are calculated by the calculation unit, and the average of the measurement data of the predetermined lot sequentially calculated by the calculation unit A value, a standard deviation σ of the measurement data of the predetermined lot calculated by the calculation unit, an initial data of the average value, and a data range prediction unit that predicts based on the initial data of the standard deviation σ,
A determination unit that determines whether the 6σ data range sequentially predicted by the data range prediction unit is out of a predetermined standard width;
A display control unit for displaying a malfunction occurrence time on the display using the value of p when the determination unit determines that the 6σ data range of the p-th lot is out of the predetermined standard width; This is in a failure occurrence prediction device characterized by the above.

  In the present invention, the average value and standard deviation σ of the measurement data for each lot are calculated based on the measurement data input by the user. The measurement data is obtained by measuring physical characteristics or electrical characteristics of a product for each lot. The physical characteristics of the product refer to, for example, the size of each product, and the electrical characteristics of the product refer to, for example, the insulation and conductivity of the wiring board of each product when the product is an electronic component. When the physical characteristics or electrical characteristics of each product are measured, it can be seen that there is variation between products.

  If the measured value of each product is a measured value of physical characteristics, for example, 20 mm, and the measured value of the electrical characteristics of each product can be expressed by a value of 20 MΩ, for example. Data is sequentially input by the user. Based on the input measurement data, an average value and standard deviation σ (hereinafter referred to as σ) of the measurement data of a predetermined lot are calculated.

  Next, the 6σ data range of each lot from the first to a predetermined number, for example, the 10th lot following the predetermined lot is predicted. The 6σ data range refers to a range between ± 3σ in a normal distribution curve representing variation in measurement data. This 6σ data range includes most of the measurement data of each product in the lot. The 6σ data range of each lot from the first to the tenth following the predetermined lot is based on the calculated average value and standard deviation σ of the measurement data of the predetermined lot, and the initial data of the average value and the initial data of the standard deviation σ. Predicted. The initial data of the average value is the average value of the measurement data of each lot up to, for example, the fourth lot from the first lot, and the initial data of the standard deviation σ is the initial data of the lot from the first lot, for example, to the fourth lot. The standard deviation σ of the measurement data is used.

  Further, the difference between the average value of the measurement data of the predetermined lot and the initial data of the average value and the difference between the standard deviation σ of the measurement data of the predetermined lot and the initial data of the standard deviation σ are detected. It is determined whether at least one of them is greater than or equal to a predetermined value, and only when it is greater than or equal to a predetermined value, the 6σ data range of each of the first to tenth lots following the predetermined lot is predicted. Good. By doing so, it is possible to predict only the 6σ data range of a lot that may cause a defect in the future.

  Next, it is determined whether or not the predicted 6σ data range is out of a predetermined standard width. If it is within a predetermined standard width, it is considered that there is no possibility that a physical or electrical defect occurs in the measurement data of the product. If even a part of the predicted 6σ data range is outside the standard range, there is a possibility that some of the products in the lot may have physical or electrical defects.

  Of the 1st to 10th lots following the predetermined lot, for example, if it is determined that a part of the 6σ data range of the 5th lot is out of the predetermined standard width, the failure occurrence time is displayed using the lot number. Is displayed. On the display, for example, a lot number that may cause a defect is displayed. Users can easily check the number of lots that are predicted to be defective in the future by looking at the information displayed on the display, and prevent the occurrence of defects before this lot is produced. can do.

  At this time, a warning may be generated. As a method for generating a warning, a warning may be displayed on the display using characters at the same time as the failure occurrence time, or a buzzer or the like may be provided in the failure occurrence predicting device to notify the user with a sound. Thus, when a warning is generated using a display, the user can easily be notified of the warning together with the prediction result. When a warning is generated using a buzzer, the user is away from the failure prediction device. Even if it is, it can be notified that there is a risk of malfunction.

  Next, when it is determined that both of the two detected differences are less than the predetermined value, the calculated average value of the measurement data of the lot immediately before the predetermined lot and the measurement data of the predetermined lot The difference between the average value and the standard deviation σ of the measurement data of the lot immediately before the predetermined lot and the standard deviation σ of the measurement data of the predetermined lot are detected.

  A warning is generated when it is determined that at least one of the two detected differences is greater than or equal to a warning generation level at which a warning should be generated. As warning generation levels, for example, a plurality of warning generation levels, for example, three warning generation levels whose risk levels are changed in stages are prepared. The warning is displayed on, for example, a display, and is displayed with a display such as high risk level, medium risk level, low risk level, etc. according to the level of risk level.

  In this way, the degree of risk changes based on the magnitude of the detected difference, and the degree of danger is displayed on the display in an easy-to-understand manner, so that the user can instantly understand the degree of danger of the warning.

  Further, when it is determined that at least one of the two detected differences is equal to or higher than a warning generation level at which a warning should be generated, past product information is displayed on the display. The past product information is, for example, in producing a product such as what number of molds that are one of the parts that make up the product was updated in the past, and when the factory that produces the product was changed. Information that affects product measurement data.

  When past product information is displayed on the display in this way, if the detected difference is greater than or equal to the warning occurrence level, the user can easily confirm whether the cause is due to past product information. Can do.

  Further, when it is determined that at least one of the two detected differences is equal to or higher than a warning generation level at which a warning is to be generated, the initial data of the average value and the initial data of the standard deviation σ are measured data of a predetermined lot. To the average value and standard deviation σ. That is, if the initial data of the average value is the average value from the first lot to the fourth lot, and the initial data of the standard deviation σ is the standard deviation σ from the first lot to the fourth lot, 3 is changed to the average value and standard deviation σ of the measurement data of a predetermined lot for which the difference is determined to be greater than or equal to the warning occurrence level.

  In this way, if it is determined that the difference from the previous lot is greater than or equal to the warning occurrence level, the initial data is changed to the average value and standard deviation σ of the measurement data of the predetermined lot, so prediction accuracy will continue to be improved in the future. The time of occurrence of a failure is predicted without deteriorating.

  According to the present invention, the average value and standard deviation σ of the measurement data of a predetermined lot are obtained, and the 6σ data range of the first to predetermined lots following the predetermined lot is sequentially estimated using these, and the pth lot When the data range is out of the predetermined standard range, the value of p is used to display the time when the trouble is expected to occur, and the user grasps the time when the trouble is expected to occur, and the product It is possible to prevent problems from occurring.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. FIG. 1 shows a configuration of a failure occurrence prediction apparatus 10 as an embodiment. This failure occurrence prediction device 10 includes a CPU (Central Processing Unit) 11, a ROM (Read Only Memory) 12, a RAM (Random Access Memory) 13, an input / output I / F (Interface) 14, and an HDD (Hard Disk). Drive) 15 and these are connected to the bus 19. The input / output I / F 14 is connected to a keyboard 16 as a user interface, a mouse 17, and a display 18.

  The CPU 11 controls the overall operation of the failure occurrence prediction device 10. The ROM 12 stores a control program and data for controlling the operation of the CPU 11. The RAM 13 functions as a working area for the CPU 11. The CPU 11 reads out the control program stored in the ROM 12 as necessary, transfers it to the RAM 13 and develops it. And CPU11 controls each part of the malfunction occurrence prediction apparatus 10 by reading and executing the control program developed by RAM13.

  The input / output I / F 14 defines a data exchange control method and the like. For example, measurement data input from the keyboard 16 passes through the bus 19 via the input / output I / F 14 and is processed by the CPU 11. The display information controlled by the CPU 11 is supplied to the display 18 via the input / output I / F 14. The display 18 includes a display device such as an LCD (Liquid Crystal Display). The HDD 15 temporarily stores measurement data taken in via the input / output I / F 14.

  Hereinafter, the operation of the CPU 11 of the failure occurrence prediction apparatus 10 will be described with reference to the drawings.

  First, the user takes out all of the products produced in each lot or some of the products produced in each lot, and measures physical characteristics or electrical characteristics according to the products. The user inputs the measured value using, for example, the keyboard 16. The inputted measurement value is taken in via the input / output I / F 14 as measurement data. The measurement data is temporarily stored in the HDD 15 via the bus 19. The CPU 11 calculates the average value and standard deviation σ of the measurement data for each lot based on the acquired measurement data. The calculation method of the average value and the standard deviation σ is stored in advance in the ROM 12, for example. At this time, the CPU 11 constitutes a calculation unit.

  Next, based on the average value and the standard deviation σ that are sequentially calculated, and the initial data of the average value and the initial data of the standard deviation σ, the CPU 11 starts from a first lot that follows a predetermined lot, for example, a lot that is currently being produced. Predict the 6σ data range of the 10th lot. Here, the initial data of the average value is the average value of the measurement data from the first lot to the fourth lot, for example, and the initial data of the standard deviation σ is, for example, the measurement from the first lot to the fourth lot The standard deviation σ of the data is used. The initial data of the average value and the initial data of the standard deviation σ are stored in advance in the RAM 13, for example.

  The CPU 11 detects the difference between the average value of the measurement data of the predetermined lot and the average value of the initial data, and the difference between the standard deviation σ of the measurement data of the predetermined lot and the standard deviation σ of the initial data. At this time, the CPU 11 constitutes a first difference detection unit.

  Next, the CPU 11 determines whether at least one of the two detected differences is equal to or greater than a predetermined value. At this time, the CPU 11 constitutes a first determination unit. Here, if the CPU 11 determines that at least one of the two detected differences is greater than or equal to a predetermined value, there is a risk that a future product will be defective.

  FIG. 2 is a diagram showing a distribution curve based on measured data from the first lot to the 200th lot, and a distribution curve based on a predicted 6σ data range of the pth lot to be produced in the future. The 6σ data range refers to a range between ± 3σ in a normal distribution curve representing variation in measurement data. This 6σ data range includes most of the measurement data of each product in the lot.

  As shown in FIG. 2, the measurement data for each lot generally changes as the lot number increases to 1, 100, 200 due to the consumption of molds and jigs used in the product. Go.

  The distribution curves from the first lot to the 200th lot are all within the standard range. The standard width refers to a criterion for determining that a product is on the pass line. For example, when measuring the physical characteristics of a product, if an error such as the length of one part of the product is up to ± 0.1 mm and an acceptable line is used, the error ± 0.1 mm is the standard width. If a part of the distribution curve of a lot exceeds this standard width, there is a possibility that a product in the lot has a defect.

  As shown in the figure, the pth lot following the 200th lot is the first lot whose predicted 6σ data range falls outside the standard range. Assuming that the currently produced lot is the 200th lot, the CPU 11 sequentially predicts the 6σ data range based on the measurement data of the 1st to 10th lots following the 200th lot, that is, the 201st to 210th lots. To go. At this time, the CPU 11 constitutes a data range prediction unit.

  Next, the CPU 11 sequentially determines whether the predicted 6σ data range exceeds the standard width as in the p-th lot shown in the drawing. As a result of the determination, the CPU repeats a series of operations when it is determined that all the 6σ data ranges of the first to tenth lots are within the standard range, and when the next lot enters the production stage. At this time, the CPU 11 constitutes a second determination unit.

  Conversely, if the CPU 11 predicts the 6σ data range of the p-th, for example, the fifth lot from the 200th lot and determines that the data range exceeds the standard width, the value of p, in this case, 5 is used to display on the display 18 the failure occurrence time when the failure is predicted to occur. At this time, the CPU 11 constitutes a first display control unit.

  If the CPU 11 determines that the 6σ data range of the 205th lot exceeds the standard width, it issues a warning. The warning is displayed on the display 18 together with the above-described malfunction occurrence time. FIG. 3 shows a display example displayed on the display 18 at this time. On the left side of the screen are a warning display and a failure occurrence time display, and on the right side of the screen, the diagram shown in FIG. 2 is displayed. At this time, the display 18 constitutes a warning generation unit, and the CPU 11 constitutes a first warning control unit.

  Next, the difference between the average value of the measurement data of the predetermined lot and the average value of the initial data and the difference between the standard deviation σ of the measurement data of the predetermined lot and the standard deviation σ of the initial data are detected, and both of the two differences are detected. A case where the predetermined value is not satisfied will be described.

  The CPU 11 calculates the average value of the measurement data of the lot immediately before the predetermined lot, the average value of the measurement data of the predetermined lot, the standard deviation σ of the measurement data of the lot immediately before the predetermined lot, and the measurement data of the predetermined lot. The difference from the standard deviation σ is detected. As shown in FIG. 4, for example, even if the difference between the initial data of the average value and the initial data of the standard deviation σ does not exceed a predetermined value, the average of the measurement data of the previous lot for some reason This is because it is considered that a difference between the value and the average value of the measurement data of the predetermined lot appears remarkably. At this time, the CPU 11 constitutes a second difference detection unit.

  Further, the CPU 11 determines whether or not at least one of the two detected differences exceeds a certain level (warning occurrence level) at which a warning is generated. The warning occurrence level is divided into, for example, three levels of “low risk”, “medium risk”, and “high risk”. The CPU 11 determines in order from “low risk” whether the detected difference is equal to or higher than each warning occurrence level. In this case, the CPU 11 constitutes a third determination unit.

  If the CPU 11 determines that at least one of the two differences is equal to or higher than the warning occurrence level, the CPU 11 then displays a level indicating the warning and the warning risk level based on the detected difference on the display 18. FIG. 6 shows a display example displayed on the display 18. As shown in FIG. 6, past product information is displayed on the display 18 together with the warning and the danger level of the warning.

  The past product information refers to, for example, the production of a product, such as the number of lots in which a mold, which is one of the parts that make up the product, was updated in the past, or when the factory that produces the product was changed. Information that affects the measurement data of a product. The reason why past product information is displayed together on the display 18 is that some change in production of the product is often one of the causes that the detected difference is significant. In this case, as shown in FIG. 6, it can be confirmed that the mold used for the product has been updated with the 101st lot which is the lot before the 102nd lot being last. Thus, it can be seen that a difference is detected in the average value of the measurement data of the 101st lot and the 102nd lot. At this time, the display 18 constitutes a warning generation unit, and the CPU 11 constitutes a second warning control unit.

  In this case, the initial data of the average value is changed by the user to the average value of the measurement data of the 102nd lot, and the initial data of the standard deviation σ is changed to the standard deviation σ of the measurement data of the 102nd lot. In this way, if it is determined that the difference from the previous lot is greater than or equal to the warning occurrence level, the initial data is changed to the average value and standard deviation σ of the measurement data of the predetermined lot, so prediction accuracy will continue to be improved in the future. The time of occurrence of a failure is predicted without deteriorating.

  These changes in initial data are input via the keyboard 16 and sent to the CPU 11 via the bus 19 via the input / output I / F 14. The CPU changes the initial data of the average value and the initial data of the standard deviation σ stored in the RAM 13 based on the information input from the keyboard 16. At this time, the CPU 11 constitutes an initial data changing unit.

  Next, the operation performed in the failure occurrence prediction apparatus 10 will be described using the flowchart of FIG.

  First, starting from step ST1, in step ST2, the user measures the physical characteristics or electrical characteristics of the nth lot currently being produced, and inputs the measured values as measurement data on the keyboard 16.

  Next, in step ST3, the CPU 11 reads out the calculation method of the average value and the standard deviation σ from the ROM 12 based on the measurement data input in step ST2, and calculates the average value Xn of the measurement data of the nth lot and the measurement data. A standard deviation σn is calculated.

  In step ST4, the CPU 11 reads the difference detection method from the ROM 12, detects | ΔXn | which is the difference between the average value Xn of the nth lot and the initial data Xa of the average value, and this | ΔXn | It is determined whether the value is greater than or equal to α, or the difference | Δσn | between the standard deviation σn of the n-th lot and the initial data σa of the standard deviation σ is detected, and whether or not | Δσn | to decide. The initial data Xa of the average value is, for example, the average value of the measurement data of the first lot to the fourth lot, and the initial data σa of the standard deviation σ is, for example, the measurement from the first lot to the fourth lot. The standard deviation σ of the data.

  If it is determined in step ST4 that | ΔXn | is greater than or equal to α or | Δσn | is greater than or equal to β, in step ST6, the CPU 11 determines the nth value based on Xa, Xn, σa, and σn. From the first to the tenth following the lot, the 6σ data range of the pth lot, that is, {Xn + (ΔXn / n) * p} −3 * {σn + (Δσn / n) * p} to {Xn + ( ΔXn / n) * p} + 3 * {σn + (Δσn / n) * p} is predicted.

  In step ST7, the CPU 11 determines whether or not the 6σ data range predicted in step ST6 is outside the standard range that causes a defect in the product. If it is determined in step ST7 that it is out of the standard range, a warning and a malfunction occurrence time are displayed on the display 18 in step ST8 (see FIG. 3). Thereafter, the process proceeds to Step 12.

  Next, in step ST4, when the CPU 11 determines that | ΔXn | is less than α and that | Δσn | is less than β, in step ST5, the CPU 11 determines the previous lot of the nth lot. It is determined whether the difference (| Xn-1−Xn |) between the average value Xn−1 of the measurement data of the n−1th lot and Xn is γ or more, or the n−1th lot It is determined whether or not the difference (| σn−1−σn |) between the standard deviation σn−1 of the measured data and σn is δ or more.

  In this case, γ and δ represent a plurality of stepwise numerical values. γ is, for example, 1/20, 1/10, 1/5 of the initial data Xa of the average value, δ is, for example, 1/20, 1/10, 1/5 of σa, and the denominator is The smaller the risk, the greater the risk. The CPU 11 determines the difference between Xn−1 and Xn, or the difference between σn−1 and σn, in order from a numerical value with a large denominator, in this case, Xa / 20 or σa / 20.

  In step ST5, the CPU 11 determines that the difference (| Xn−Xn−1 |) of the average values is less than γ and that the difference (| σn−σn−1 |) of the standard deviation σ is not less than δ. In step ST9, the CPU 11 displays the calculated average value and standard deviation σ, and displays on the display 18 that the measured data of the product is within the standard range indicating that it is in the pass line.

  FIG. 5 is a diagram showing a histogram based on the average value and standard deviation σ of the lot and the measurement data of the products in the lot. If the CPU 11 determines in step ST7 that the 6σ data range predicted in step ST6 is not outside the standard range, the process similarly proceeds to step ST9 and the screen of FIG. After this, the process proceeds to step ST12.

  If the CPU 11 determines in step ST5 that the average value difference is γ or more or the standard deviation σ difference is δ or more, a warning and past product information are displayed on the display 18 in step ST10. indicate. As described above, the screen shown in FIG. 6 is displayed on the display 18 at this time. At this time, the difference is detected using the average value as a guideline. Since the difference in this case exceeds Xa / 5, the degree of danger is displayed as “large”. Thereafter, a display example showing the transition of the average value of the measurement data for each lot up to the current lot shown in FIG. 7 may be displayed on the display 18.

  The user looks at the screens of FIG. 6 and FIG. 7 displayed in step ST10 and confirms past product information, and then in step ST11, the initial value Xa of the average value is the average value of the nth lot Xn. The initial data σa of the standard deviation σ is changed to σn that is the standard deviation σ of the nth lot. Next, the process proceeds to step ST12.

  In step ST12, a form for inputting measurement data is displayed on the display 18 when the (n + 1) th lot, which is the next lot after the nth lot, has entered the production stage, and hereinafter, from step ST2 to step ST13. A series of operations is repeated.

  As described above, according to the failure occurrence prediction apparatus 10 shown in FIG. 1, the average value and the standard deviation σ of the measurement data of the nth lot are obtained, and using these, for example, from the 1st to the 10th from the nth lot. The 6σ data range of the first lot is predicted, and when the 6σ data range of the pth lot is out of the standard width, the occurrence time of the defect is displayed on the display 18 using the value of p to notify the user. The user can prevent the product from malfunctioning.

  In the above-described embodiment, when a warning is generated, a warning is displayed on the display 18. In addition to the display 18, a warning by a sound using a buzzer or the like, a light, or the like is used. It may be a warning by light emission.

  In the above embodiment, when the initial data of the average value and the initial data of the standard deviation σ are changed, it is changed by the user. However, in step ST5 of the flowchart of FIG. When the CPU 11 determines that the difference (| Xn-1−Xn |) is greater than or equal to γ, or the difference (| σn−1−σn |) of the standard deviation σ is greater than or equal to δ, the nth lot The average value may be automatically changed to the initial data of the average value, and the standard deviation σ of the nth lot may be automatically changed to the initial data of the standard deviation σ.

  Further, in the above-described embodiment, when a warning is generated in step ST8 of FIG. 8, as shown in FIG. 3, the lot that is predicted to be defective is displayed as the pth from the current lot. Alternatively, the lot number may be used for display. For example, if the current lot is the 200th, it is displayed on the screen shown in FIG. 3 that the lot number 200 + p may be out of the standard range.

  Further, in the above-described embodiment, as shown in FIGS. 6 and 7 in step ST10 of FIG. 8, a display example based on the average value of the measurement data of the predetermined lot is displayed on the display 18. When it is determined in step ST5 that the difference (| σn−σn−1 |) of the standard deviation σ is equal to or larger than δ, a display example based on the standard deviation σ may be displayed.

  The present invention predicts a time when a defect occurs in a product produced for each lot, and can be used when a product is manufactured in a factory or the like.

It is a figure which shows the structure of the malfunction occurrence prediction apparatus as embodiment. It is a figure which shows the distribution curve based on the measured data for every lot, and the distribution curve based on the estimated 6-sigma data range of the p-th lot. It is the example of a display which displayed the warning and the predicted malfunction occurrence time. It is the example of a display which displayed that the difference of the average value of a predetermined lot and the average value of the last lot is more than a predetermined value. It is the example of a display which displayed the average value and standard deviation (sigma) of a predetermined lot, and the histogram of measurement data. It is the example of a display which displayed the warning, the risk, and the past product information. It is the example of a display which displayed the change of the data average value, and the past product information. It is a flowchart which shows the operation | movement performed with a malfunction occurrence prediction apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 ... Failure occurrence prediction device, 11 ... CPU, 12 ... ROM, 13 ... RAM, 14 ... Input / output I / F, 15 ... HDD, 16 ... Keyboard, 17 ... Mouse, 18 ... Display, 19 ... Bus

Claims (8)

A measurement data input unit for inputting measurement data obtained by measuring physical characteristics or electrical characteristics of a product for each lot;
Based on the measurement data input in the measurement data input unit, a calculation unit that calculates the average value and standard deviation σ of the measurement data for each lot;
The 6σ data range from the first lot to the predetermined number of lots following the predetermined lot for which the average value and standard deviation σ of the measurement data are calculated by the calculation unit, and the average of the measurement data of the predetermined lot sequentially calculated by the calculation unit A value, a standard deviation σ of the measurement data of the predetermined lot calculated by the calculation unit, an initial data of the average value, and a data range prediction unit that predicts based on the initial data of the standard deviation σ,
A determination unit that determines whether the 6σ data range sequentially predicted by the data range prediction unit is out of a predetermined standard width;
A display control unit for displaying a malfunction occurrence time on the display using the value of p when the determination unit determines that the 6σ data range of the p-th lot is out of the predetermined standard width; A failure occurrence prediction device characterized by the above. The determination unit further includes a warning control unit that generates a warning from the warning generation unit when it is determined that the 6σ data range of the p-th lot is out of the predetermined standard range. The malfunction occurrence prediction apparatus according to claim 1. A measurement data input unit for inputting measurement data obtained by measuring physical characteristics or electrical characteristics of a product for each lot;
Based on the measurement data input in the measurement data input unit, a calculation unit that calculates the average value and standard deviation σ of the measurement data for each lot;
The difference between the average value of the measurement data of the predetermined lot calculated by the calculation unit and the initial data of the average value, and the standard deviation σ of the measurement data of the predetermined lot calculated by the calculation unit and the initial value of the standard deviation σ A first difference detection unit for detecting a difference from data;
A first determination unit that determines whether at least one of the two differences detected by the first difference detection unit is greater than or equal to a predetermined value;
When the first determination unit determines that at least one of the two differences detected by the first difference detection unit is equal to or greater than a predetermined value, the first to the predetermined number of times following the predetermined lot 6 sigma data range of the lots, the average value of the measurement data of the predetermined lot sequentially calculated by the calculation unit, the standard deviation σ of the measurement data of the predetermined lot calculated by the calculation unit, the initial data of the average value and the above A data range prediction unit for predicting based on the initial data of the standard deviation σ;
A second determination unit that determines whether the 6σ data range sequentially predicted by the data range prediction unit is out of a predetermined standard range;
When it is determined by the second determination unit that the 6σ data range of the p-th lot is out of the predetermined standard range, the first occurrence time is displayed on the display using the value of p. A failure occurrence predicting device comprising: a display control unit. A first warning control unit that generates a warning from the warning generation unit when the second determination unit determines that the 6σ data range of the p-th lot is out of the predetermined standard width; The failure occurrence prediction device according to claim 3, wherein the failure occurrence prediction device is provided. When the first determination unit determines that both of the two differences detected by the first difference detection unit are less than a predetermined value, the average of the measurement data of the predetermined lot calculated by the calculation unit The difference between the value and the average value of the measurement data of the lot immediately before the predetermined lot, the standard deviation σ of the measurement data of the predetermined lot calculated by the calculation unit, and the predetermined lot calculated by the calculation unit A second difference detector for detecting a difference from the standard deviation σ of the measurement data of the previous lot;
A third determination unit that determines whether or not the difference detected by the second difference detection unit is equal to or higher than a warning occurrence level;
When the third determination unit determines that at least one of the two differences detected by the second difference detection unit is equal to or higher than a warning occurrence level, a second warning is generated from the warning generation unit. Two warning control units;
When the third determination unit determines that at least one of the two differences detected by the second difference detection unit is equal to or higher than the warning occurrence level, the past product information is displayed on the display. The defect occurrence prediction apparatus according to claim 3, further comprising a second display control unit. The third determination unit prepares a plurality of warning generation levels that change stepwise as the warning generation level, and the difference detected by the second difference detection unit is equal to or higher than each warning generation level. Determine if there is,
The said 2nd warning control part changes the content of the warning generate | occur | produced from the said warning generation part according to the judgment result with respect to these warning generation levels of the said 3rd judgment part. 5. A failure occurrence prediction device according to 5. When it is determined by the third determination unit that at least one of the two differences detected by the second difference detection unit is on the warning generation level, the initial data of the average value is used as the predetermined lot. 6. The defect according to claim 5, further comprising an initial data changing unit that changes the average value of the measurement data of the standard deviation σ to the standard deviation σ of the measurement data of the predetermined lot. Occurrence prediction device. A measurement data input step for inputting measurement data obtained by measuring physical characteristics or electrical characteristics of a product for each lot;
Based on the measurement data input in the measurement data input step, a calculation step for calculating the average value and standard deviation σ of the measurement data for each lot;
The 6σ data range of the first lot to the predetermined number of lots following the predetermined lot for which the average value of the measurement data and the standard deviation σ were calculated in the calculation step, and the average of the measurement data of the predetermined lot sequentially calculated in the calculation step A data range prediction step for predicting a value based on the standard deviation σ of the measurement data of the predetermined lot calculated in the calculation step, the initial data of the average value, and the initial data of the standard deviation σ,
A determination step of determining whether or not the 6σ data range sequentially predicted in the data range prediction step is out of a predetermined standard width;
A display control step of displaying a failure occurrence time on the display using the value of p when it is determined in the determination step that the 6σ data range of the p-th lot is out of the predetermined standard width; A failure occurrence prediction method characterized by the above.

Images