Manufacturing defect factor searching method and manufacturing defect factor searching apparatus

Patent metadata

Field	Value
Publication number	US-11592807-B2
Application number	US-202117220145-A
Country	US
Kind code	B2
Filing date	Apr 1, 2021
Priority date	Apr 8, 2020
Publication date	Feb 28, 2023
Grant date	Feb 28, 2023

Abstract

A manufacturing defect factor searching method includes: classifying manufacturing monitoring data into a set of non-defective products having an inspection result indicating a non-defective product and a set of defective products having the inspection result indicating a defective product, in accordance with a correspondence relationship between the manufacturing monitoring data and product inspection data indicating the inspection result of the product manufactured in the manufacturing line, the manufacturing monitoring data being collected from a manufacturing line of a product and being multivariate; estimating, for each item of the manufacturing monitoring data, a mixture distribution function approximating to a statistical distribution of each of the set of non-defective products and the set of defective products; resolving the mixture distribution function into components; and generating a list of items including a resolved component having a correlation with a manufacturing quality defect from among items of the manufacturing monitoring data.

First claim

What is claimed is:

1. A manufacturing defect factor searching method comprising:
classifying manufacturing monitoring data into a set of non-defective products in which an inspection result indicates a non-defective product and a set of defective products in which the inspection result indicates a defective product, in accordance with a correspondence relationship between the manufacturing monitoring data and product inspection data, the manufacturing monitoring data being collected from a manufacturing line of a product and being multivariate, the product inspection data indicating the inspection result of the product manufactured in the manufacturing line;
estimating a mixture distribution function for each item of the manufacturing monitoring data, the mixture distribution function approximating to a statistical distribution of each of the set of non-defective products and the set of defective products;
resolving the mixture distribution function into components;
generating a list of items including a resolved component having a correlation with a manufacturing quality defect from among items of the manufacturing monitoring data; and
calculating a manufacturing quality defective rate function for each of the components into which the mixture distribution function has been resolved, the manufacturing quality defective rate function defining a manufacturing quality detective rate with respect to a value of each of the items of the manufacturing monitoring data, wherein
in the generating of the list, a list of the items of the manufacturing monitoring data is generated in accordance with the manufacturing quality defective rate; and
calculating a defect sensitivity index of a resolved component of a statistical distribution of the items of the manufacturing monitoring data, by using the manufacturing quality defective rate function, wherein
in the generating of the list, values of the defect sensitivity index are sorted in descending order, and the list of the items of the manufacturing monitoring data is generated.

2. The manufacturing defect factor searching method according to claim 1 , wherein
in the calculating of the defect sensitivity index, at least one of:
a defective rate function correlation between a value of the manufacturing quality defective rate function and a value of each of the items of the manufacturing monitoring data;
a maximum value of the manufacturing quality defective rate function in a confidence interval of the statistical distribution;
a defect prediction accuracy obtained from genuineness of a defect of the product included in a defect prediction interval having a great value of the manufacturing quality defective rate function and a number of prediction positive/negative sections;
a defect prediction recall ratio obtained from the genuineness of the defect of the product included in the defect prediction interval having a great value of the manufacturing quality defective rate function and the number of prediction positive/negative sections; and
an F-value serving as a harmonic mean of the defect prediction accuracy and the defect prediction recall ratio
is calculated as the defect sensitivity index.

3. The manufacturing defect factor searching method according to claim 1 wherein
the manufacturing monitoring data includes manufacturing attribute data that includes a plurality of sets of manufacturing elements that are assumed to have compatibility for playing an identical role in the manufacturing line,
a defective product responsibility rate of an element of an item of the manufacturing attribute data is calculated for each of the components into which the mixture distribution function has been resolved, and
in the generating of the list, a list of a plurality of the items of the manufacturing attribute data is generated in accordance with the defective product responsibility rate.

4. The manufacturing defect factor searching method according to claim 3 , wherein
the plurality of the items of the manufacturing attribute data includes at least one of a supply source, a manufacturing device, an inspection device, and an operator.

5. The manufacturing defect factor searching method according to claim 1 further comprising:
displaying the list of the items of the manufacturing monitoring data, the list being generated in the generating of the list.

6. A manufacturing defect factor searching apparatus comprising:
a preprocessing unit that classifies manufacturing monitoring data into a set of non-defective products in which an inspection result indicates a non-defective product and a set of defective products in which the inspection result indicates a defective product, in accordance with a correspondence relationship between the manufacturing monitoring data and product inspection data, the manufacturing monitoring data being collected from a manufacturing line of a product and being multivariate, the product inspection data indicating the inspection result of the product manufactured in the manufacturing line;
a mixture distribution function estimation unit that estimates a mixture distribution function for each item of the manufacturing monitoring data, the mixture distribution function approximating to a statistical distribution of each of the set of non-defective products and the set of defective products;
a component resolution unit that resolves the mixture distribution function into components; and
a screen generation unit that generates a list of items including a resolved component having a correlation with a manufacturing quality defect from among items of the manufacturing monitoring data;
a defect sensitivity index calculation unit that calculates a manufacturing quality defective rate function for each of the components into which the mixture distribution function has been resolved, the manufacturing quality defective rate function defining a manufacturing quality detective rate with respect to a value of each of the items of the manufacturing monitoring data, wherein
the screen generation unit generates a list of the items of the manufacturing monitoring data in accordance with the manufacturing quality defective rate;
wherein:
the defect sensitivity index calculation unit calculates a defect sensitivity index of a resolved component of a statistical distribution of the items of the manufacturing monitoring data, by using the manufacturing quality defective rate function, and
the screen generation unit sorts values of the defect sensitivity index in descending order, and generates the list of the items of the manufacturing monitoring data.

7. The manufacturing defect factor searching apparatus according to claim 6 , wherein
the defect sensitivity index calculation unit calculates, as the defect sensitivity index, at least one of:
a defective rate function correlation between a value of the manufacturing quality defective rate function and a value of each of the items of the manufacturing monitoring data;
a maximum value of the manufacturing quality defective rate function in a confidence interval of the statistical distribution;
a defect prediction accuracy obtained from genuineness of a defect of the product included in a defect prediction interval having a great value of the manufacturing quality defective rate function and a number of prediction positive/negative sections;
a defect prediction recall ratio obtained from the genuineness of the defect of the product included in the defect prediction interval having a great value of the manufacturing quality defective rate function and the number of prediction positive/negative sections; and
an F-value serving as a harmonic mean of the defect prediction accuracy and the defect prediction recall ratio.

8. The manufacturing de

Assignees

• Hitachi Ltd

Inventors

• Tamaki Kenji

Classifications

• G05B19/4183

  characterised by data acquisition, e.g. workpiece identification · CPC title

• G06V10/44

  Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components · CPC title

• Y02P90/02

  Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS] · CPC title

• G05B19/41875 Primary

  characterised by quality surveillance of production · CPC title

• G06V20/52

  Surveillance or monitoring of activities, e.g. for recognising suspicious objects (recognising microscopic objects G06V20/69) · CPC title