JPH0727672A - Method and apparatus for predicting defective state of article - Google Patents

Abstract

PURPOSE:To provide an apparatus which can predict a defective state of an article in future from quality characteristics of a market product which is normally functioning without using trouble data. CONSTITUTION:The apparatus for predicting a defective state of an article comprises a first input means for sampling a characteristic value for representing quality of an article from a plurality of articles at an initial time of starting to use to input the data, a first calculating means for calculating a distributed state of the value of the article in future from the data input from the input means, a second input means for setting or estimating the value corresponding to an allowable limit required for a market regarding the quality of the article to input the data, a comparing means for comparing the distributed state of the value of the article in future calculated by the first calculating means with the characteristic value corresponding to an allowable limit input by the second input means, and a second calculating means for calculating a defective rate of the article in future based on comparison by the comparing means.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and apparatus for predicting a defective state of an article.

[0002]

2. Description of the Related Art As a method for predicting the life of structural members,
For example, as disclosed in Japanese Patent Application Laid-Open No. 4-355338, a method of predicting the life of a structural member based on a calculation model from a crack existing on the surface of the structural member has been proposed.

By the way, a large number of parts (articles) are used in automobiles mass-produced in factories. As a method for obtaining the failure probability of these used parts, based on failure data obtained from recovered failed products. Methods for estimating future failure probabilities are generally known.

In the conventional method of estimating the failure probability, (1) a method of obtaining a failure histogram for each mileage of an automobile, (2) a method of obtaining a distribution function of each of data on the occurrence of a failure and (3) ) A method that uses the number of failures is adopted.

[0005]

However, the above-mentioned conventional method of predicting a defective state of an article is premised on failure data. For example, even in the method of obtaining the distribution function, in the conventional method, the distribution function and the number of failed vehicles are used. Therefore, if something goes wrong,
No action can be taken until the defect becomes apparent, which will increase the loss for both users and manufacturers.

Further, in general, when introducing a new technology into a mass-produced product, since the usage pattern and load in the market cannot be predicted completely, mass-production is started with the quality being closer to the over-quality side. Although we try to narrow down to the proper quality while observing the reaction of, the conventional method of predicting the defective state of an article cannot calculate the defective rate unless a defect occurs, so the life of the article cannot be estimated. However, it was difficult to make the quality of the articles converge to the proper quality. Therefore, there is a possibility that the article becomes unnecessarily high quality and therefore expensive, and imposes an unnecessary economic burden on the user.

In view of the above-mentioned circumstances, it is an object of the present invention to provide a method and apparatus capable of predicting a defective state of an article in the future without using failure data.

[0008]

According to the method of predicting a defective state of an article according to the present invention, a characteristic value representing the quality of the article is sampled from a plurality of articles at the beginning of use, the data is input, and the sampling is performed. From the data representing the distribution of characteristic values of a plurality of articles, calculate the distribution state of the characteristic values of the article in the future, and set or estimate the characteristic value corresponding to the allowable limit required by the market for the quality of the article. Inputting the data, comparing the distribution state of the calculated characteristic value of the article in the future with the characteristic value corresponding to the allowable limit, and calculating the defective rate of the article in the future. Is.

Further, as shown in FIG. 1, the apparatus for predicting a defective state of an article according to the present invention is characterized in that, as shown in FIG. Input means and first calculation means for calculating the distribution state of the characteristic value of the article in the future from the data input by the first input means, and the allowable limit required by the market regarding the quality of the article. The second to set or estimate the corresponding characteristic value and input the data
And a comparing means for comparing the distribution state of the characteristic value of the article in the future calculated by the first calculating means with the characteristic value corresponding to the allowable limit input by the second input means. And a second calculation means for calculating a future defective rate of the article based on the comparison by the comparison means.

[0010]

According to the present invention, characteristic values representing the quality of the articles are sampled from a plurality of articles in the early stage of use, that is, from marketed articles that are functioning normally in the initial stage after introduction of mass production. , The distribution state of the characteristic values of the above-mentioned articles in the future is calculated from the distribution state of the sampled characteristic values, and the failure rate in the future is predicted by this, so measures should be taken before the defects become apparent. It will be possible. Therefore, the loss of both the user and the manufacturer can be reduced.

Further, there is also an advantage that it is possible to speed up the manufacturing period of an article having an appropriate quality by predicting the future defective rate from the substitute characteristic level.

[0012]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS First, the basic concept of a method for predicting a defective state of an article, which is executed by using the apparatus according to the present invention, will be described with reference to the flowcharts shown in FIGS. In addition,
The target article is an automobile part whose quality deteriorates as the traveling distance of the automobile increases.

(1) A characteristic value representing the quality of the article is sampled from a plurality of articles at the beginning of use and the data is input to a microcomputer to draw a scatter diagram.

(2) The regression line is drawn by approximating the data drawn in the scatter diagram by the method of least squares or the like, and the deterioration function is obtained.

(3) The distribution of the characteristic values of the article is obtained from the data drawn on the scatter diagram.

(4) From the deterioration function line and the distribution state of characteristic values, the distribution state of future characteristic values of the article is calculated.

(5) Set or estimate a characteristic value corresponding to the permissible limit required by the market for the quality of the article, and input the data.

(6) Comparing the distribution state of the characteristic value of the article in the future with the characteristic value corresponding to the allowable limit required by the market, and calculating the defect rate based on the distribution portion exceeding the allowable limit value (prediction) To do.

Next, the case where the object is an oil seal made of rubber which is interposed on the mating surface of the engine will be described in detail with reference to FIG. If the oil seal is used for a long period of time, it becomes hard due to thermal deterioration and cannot follow the behavior of the mating surfaces, resulting in oil leakage. Therefore, the problem phenomenon is oil leakage, and the substitute characteristic representing the quality is the hardness of the oil seal.

In FIG. 4, the deterioration function line obtained from the data obtained from the n market samples is the equation y = 60 +
It is a straight line represented by 4 × x. However, x represents a traveling distance (10,000 km), y represents a quality characteristic (hardness), and the initial characteristic value is y = 60. Assuming that the distribution is a normal distribution determined by the average value and the variance, the average value of the distribution at mileage of 20,000 km is 68
When the standard deviation σ is set to 2 and the distribution state at a mileage of 50,000 km is predicted, the average value of the distribution is 80 and the standard deviation σ is 5
Became. Therefore, if the characteristic value corresponding to the allowable limit required by the market is set or estimated to be y = 90, the defective rate at the mileage of 50,000 km (the portion of the distribution that exceeds the allowable limit line) is predicted to be 2.275%. We were able to.

In the above example, the hardness is selected as a substitute characteristic representing the quality of the oil seal for sampling, but a practical characteristic (for example, oil leakage evaluation point) may be selected instead of the substitute characteristic. In addition, the deterioration function line may be a hyperbola and the distribution may be a Poisson distribution.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of a defective state prediction device for articles according to the present invention.

FIG. 2 is a first half of a flow chart for explaining a defective state prediction method for an article according to the present invention.

FIG. 3 is a second half of a flowchart for explaining a defective state prediction method for an article according to the present invention.

FIG. 4 is a diagram specifically illustrating a defective state prediction method for an article according to the present invention.

Claims (2)

[Claims] 1. A characteristic value representing the quality of the article is sampled from a plurality of articles at the initial stage of use and the data is input, and the data representing the distribution state of the characteristic values of the plurality of articles sampled is used in the future. The distribution state of the characteristic value in, the characteristic value of the article in the future calculated and set or estimated the characteristic value corresponding to the allowable limit required by the market for the quality of the article The method of predicting a defective state of an article, comprising: calculating a defective rate of the article in the future by comparing a distribution state of the item with a characteristic value corresponding to the allowable limit. 2. A first input means for sampling a characteristic value representing the quality of the article from a plurality of articles in the initial stage of use and inputting the data, and data inputted by the first input means. A first calculation means for calculating a distribution state of the characteristic values of the article in the future, and a second computing means for setting or estimating the characteristic value corresponding to the allowable limit required by the market for the quality of the article and inputting the data. Input means, and comparison means for comparing the distribution state of the future characteristic values of the article calculated by the first calculation means with the characteristic value corresponding to the allowable limit input by the second input means, And a second calculation means for calculating a future defective rate of the article based on the comparison by the comparison means.