Disclosure of Invention
Aiming at the problems, the invention provides a soft failure prediction method based on ICT and FCT.
The technical scheme provided by the invention is as follows:
the invention provides a soft failure prediction method based on ICT and FCT, which comprises the following steps:
Step S1, screening out important attention test items from ICT and FCT test items of products, and collecting ICT and FCT test data of products which are rejected because of the important attention test item related problems, thereby forming an abnormal data set;
S2, acquiring ICT and FCT test data of the products which are not returned by customers in the same batch according to the production time of the returned products, so as to form a normal data set;
s3, constructing a soft failure prediction model for predicting whether the product has soft failure according to ICT and FCT test data of the product based on the abnormal data set and the normal data set;
And S4, analyzing ICT and FCT test data of the product in real time according to the soft failure prediction model, so as to predict whether the soft failure exists in the product.
In the soft failure prediction method of the present invention, step S3 is performed by using a machine learning algorithm.
In the soft failure prediction method of the present invention, the step S3 includes the following steps:
step S3.1, carrying out normalization processing on all ICT and FCT test data in a union set of a normal data set and an abnormal data set, wherein the normalization processing algorithm is as follows:
kij＝|(Tij-μj)/σj|;
Wherein Tij represents the jth ICT and FCT test data of the product of the union of the normal data set and the abnormal data set;
muj represents the mean of the jth ICT and FCT test data for all products in the normal dataset;
σj represents the standard deviation of the jth ICT and FCT test data for all products in the normal dataset;
The total number of products in the union of the normal data set and the abnormal data set is recorded as n, the total item number of ICT and FCT test data of the products is recorded as m, 1 st item-mth item ICT and FCT test data after normalization treatment of the 1 st product, 1 st item-mth item ICT and FCT test data after normalization treatment of the 2 nd product, and 1 st item-mth item ICT and FCT test data after normalization treatment of the nth product are adopted to construct Anm, wherein,
S3.2, respectively calculating K values of all products in a union set of the normal data set and the abnormal data set; the method comprises the steps of marking a K value of an i-th product in a combination of a normal data set and an abnormal data set as Ki, wherein the Ki＝max(ki1,ki2,…,kim is taken as an abscissa, respectively taking a ratio of the number of products to be processed, which corresponds to the K value, in the total number of products to be processed, and a ratio of the number of products to be processed, which corresponds to the K value, in the total number of products to be processed, which are not to be processed, as an ordinate, establishing a coordinate system and drawing a corresponding curve, so as to obtain a product number ratio-K value curve to be processed and a product number ratio-K value curve to be processed;
step S3.3, defining the products with the absolute value of K value larger than the absolute value of K' in the abnormal data set as soft failure products, extracting ICT and FCT test data of the soft failure products to form a soft failure product data set, constructing Bn(m+1),
Where y1 represents the value of whether the 1 st product in the union of the normal and abnormal data sets is a soft-disabled product, y2 represents the value of whether the 2 nd product in the union of the normal and abnormal data sets is a soft-disabled product, yn represents the value of whether the n-th product in the union of the normal and abnormal data sets is a soft-disabled product;
When any one of y1,y2,…,yn is 0, the corresponding product is not a soft failure product, and when any one of y1,y2,…,yn is 1, the corresponding product is a soft failure product;
and S3.4, using the first m columns of Bn(m+1) as independent variables and the (m+1) th column as dependent variables, and analyzing by using a machine learning classification algorithm to construct and obtain a soft failure prediction model.
In the soft failure prediction method, the step S4 further comprises the steps of carrying out normalization processing on the ICT and FCT test data of the product in real time, and then inputting the data as independent variables into a soft failure prediction model to obtain a prediction result and outputting the prediction result.
The soft failure prediction method based on ICT and FCT is constructed, soft failure prediction is realized by combining ICT/FCT test results with a machine learning algorithm, influence caused by deflection in an acceptable range of the test results due to external force is reduced, and the passenger withdrawal rate is reduced. The soft failure prediction method based on ICT and FCT is novel in design and high in practicability.
Detailed Description
The invention aims to solve the technical problems that in the current manufacturing process of electronic products, the test results of ICT (on-line test)/FCT (functional circuit test) do not consider the deviation within the acceptable range of the test results caused by external force, and the deviation can cause the reduction of the reliability of the products, and the product customer return, the rising of repair cost and the reputation loss are easy to cause. The invention provides a soft failure prediction method based on ICT and FCT, which utilizes ICT/FCT test results to combine with machine learning algorithm to realize soft failure pre-judgment, reduces influence caused by deviation in the acceptable range of the test results caused by external force and reduces the passenger withdrawal rate.
In order to make the technical objects, technical solutions and technical effects of the present invention more apparent, so as to facilitate understanding and implementation of the present invention by those skilled in the art, the present invention will be described in further detail with reference to the accompanying drawings and specific examples.
Referring to fig. 1, fig. 1 shows a flowchart of a soft failure prediction method based on ICT and FCT according to a preferred embodiment of the present invention, and the present invention proposes a soft failure prediction method based on ICT and FCT, including the following steps:
Step S1, screening out important attention test items from ICT and FCT test items of products, and collecting ICT and FCT test data of products which are rejected because of the important attention test item related problems, thereby forming an abnormal data set;
S2, acquiring ICT and FCT test data of the products which are not returned by customers in the same batch according to the production time of the returned products, so as to form a normal data set;
here, the same batch of products refers to the same batch of products, the same production line, and the same shift.
S3, constructing a soft failure prediction model for predicting whether the product has soft failure according to ICT and FCT test data of the product based on the abnormal data set and the normal data set;
here, this step is performed by employing a machine learning algorithm.
Step S3 comprises the steps of:
step S3.1, carrying out normalization processing on all ICT and FCT test data in a union set of a normal data set and an abnormal data set, wherein the normalization processing algorithm is as follows:
kij＝|(Tij-μj)/σj|;
Wherein Tij represents the jth ICT and FCT test data of the product of the union of the normal data set and the abnormal data set;
muj represents the mean of the jth ICT and FCT test data for all products in the normal dataset;
σj represents the standard deviation of the jth ICT and FCT test data for all products in the normal dataset;
The total number of products in the union of the normal data set and the abnormal data set is recorded as n, the total item number of ICT and FCT test data of the products is recorded as m, 1 st item-mth item ICT and FCT test data after normalization treatment of the 1 st product, 1 st item-mth item ICT and FCT test data after normalization treatment of the 2 nd product, and 1 st item-mth item ICT and FCT test data after normalization treatment of the nth product are adopted to construct Anm, wherein,
S3.2, respectively calculating K values of all products in a union set of the normal data set and the abnormal data set; the method comprises the steps of marking the K value of an i-th product in a combination of a normal data set and an abnormal data set as Ki, wherein Ki＝max(ki1,ki2,…,kim) taking the K value as an abscissa, respectively taking the ratio of the number of products to be processed corresponding to the K value in the total number of products to be processed and the ratio of the number of products to be processed corresponding to the K value in the total number of products to be processed not to be processed as an ordinate, establishing a coordinate system and drawing a corresponding curve, thereby obtaining a product number ratio-K value curve to be processed and a product number ratio-K value curve to be processed not to be processed;
step S3.3, defining the products with the absolute value of K value larger than the absolute value of K' in the abnormal data set as soft failure products, extracting ICT and FCT test data of the soft failure products to form a soft failure product data set, constructing Bn(m+1),
Where y1 represents the value of whether the 1 st product in the union of the normal and abnormal data sets is a soft-disabled product, y2 represents the value of whether the 2 nd product in the union of the normal and abnormal data sets is a soft-disabled product, yn represents the value of whether the n-th product in the union of the normal and abnormal data sets is a soft-disabled product;
When any one of y1,y2,…,yn is 0, the corresponding product is not a soft failure product, and when any one of y1,y2,…,yn is 1, the corresponding product is a soft failure product;
and S3.4, using the first m columns of Bn(m+1) as independent variables and the (m+1) th column as dependent variables, and analyzing by using a machine learning classification algorithm to construct and obtain a soft failure prediction model.
And S4, analyzing ICT and FCT test data of the product in real time according to the soft failure prediction model, so as to predict whether the soft failure exists in the product.
And step S4, carrying out normalization processing on the ICT and FCT test data of the product in real time, and then inputting the data as independent variables into a soft failure prediction model to obtain a prediction result and outputting the prediction result.
In the technical scheme, soft failure means that the product has external force damage (such as electric damage, vibration damage and the like) before shipment, but the electrical property and the function of the product are not affected, and the product can be normally shipped through ICT and FCT tests.
In the use process of products with soft failure after shipment, electrical performance, functional defects and brain-maintaining patterns can be exposed in advance, so that customers can get away.
Products with soft failures can have shifts or mutations in some ICT and FCT test data.
In the above technical solutions, the product focus test items are mainly ICT and FCT test items of semiconductor devices, and mainly include analog quantity tests of semiconductor devices, such as voltage tests, current tests, capacitive reactance tests, and the like.
It will be understood that modifications and variations will be apparent to those skilled in the art from the foregoing description, and it is intended that all such modifications and variations be included within the scope of the following claims.