Search

CN-122029494-A - Method for selecting a computer-aided analysis model

CN122029494ACN 122029494 ACN122029494 ACN 122029494ACN-122029494-A

Abstract

A method for selecting a computer-aided analysis model for use in analyzing optical anomalies of a product produced on a production line is proposed, having the steps of providing at least one computer-aided analysis model, ascertaining at least one characteristic and/or at least one production parameter of the product, taking into account the at least one ascertained characteristic and/or at least one ascertained production parameter of the product, selecting an analysis model associated with a product type from a plurality of analysis models associated with the product type, and checking the product using the selected analysis model.

Inventors

  • S. Rezapurlakani
  • D. Shar

Assignees

  • 西门子股份公司

Dates

Publication Date
20260512
Application Date
20240911
Priority Date
20230922

Claims (11)

  1. 1. A method for selecting a computer-aided analysis model (34, 36, 38) for use in analyzing optical anomalies of a product produced on a production line (2), the method having the steps of: providing at least one computer-aided analysis model (34, 36, 38), Determining at least one characteristic and/or at least one production parameter of the product, Selecting an analytical model (34, 36, 38) associated with a product type from a plurality of analytical models (34, 36, 38) respectively associated with product types (4, 6, 8) taking into account at least one determined characteristic and/or at least one determined production parameter of the product, -Checking the product using the selected analytical model (34, 36, 38).
  2. 2. Method according to claim 1, characterized in that a non-optical method is used in determining at least one characteristic of the product and/or the at least one production parameter.
  3. 3. The method of claim 1, wherein one or more of the dimensions of the product, the weight of the product, the shape of the product, the color of the product, and the temperature of the product are determined when determining at least one characteristic of the product.
  4. 4. A method according to any one of claims 1 or 3, characterized in that the speed and/or the production temperature of the conveyor belt (12) is determined when determining at least one production parameter.
  5. 5. Method according to any of claims 2 to 4, characterized in that an identification code of the product is determined in addition to the determined characteristics of the product and/or in addition to the production parameters.
  6. 6. Method according to any of the preceding claims, characterized in that a binary decision tree is used in the selection of the analytical model (34, 36, 38), according to which a plurality of characteristics and/or a plurality of production parameters of the product to be inspected are taken into account in the selection of the analytical model (34, 36, 38).
  7. 7. The method according to any one of claims 1 to 5, characterized in that association rules respectively associating an analytical model with each combination of values of a plurality of input parameters are considered in selecting the analytical model (34, 36, 38).
  8. 8. The method according to any of the preceding claims, characterized in that a plurality of product types (4, 6, 8) are associated with the same analytical model.
  9. 9. The method according to any of the preceding claims, wherein selecting an analytical model (34, 36, 38) associated with a product type is performed by: Recording an image (56, 58) of the product, -Inputting said image (56, 58) and said at least one characteristic and/or said at least one production parameter into an analytical model, wherein said analytical model has been trained beforehand with structurally identical data.
  10. 10. A computer program comprising instructions which, when executed by a computer, cause the computer to perform the method of any one of the preceding claims.
  11. 11. A machine-readable storage medium on which a computer program according to the preceding claim is stored.

Description

Method for selecting a computer-aided analysis model Technical Field The invention relates to a method for selecting a computer-aided analysis model for use in analyzing optical anomalies of a product produced on a production line. Background Such computer-aided analysis models are used, for example, when identifying visual anomalies by means of a visual quality control system. What can happen here is that the production line or production line to be monitored is operated with a rapid product change. In this case, different products are thus manufactured by the same production line in different time periods. Accordingly, it can then be necessary to adapt the analytical model used to the respective product. Depending on the currently produced product, different optical parameters can indicate defects, but are carefree in terms of the quality of the inspected product. Quality control systems that identify abnormal data are typically located at the end of one or more process lines. The anomalies can be different for each product. For example, in the case of bakery applications, different optical criteria can indicate quality defects for different products (e.g., farinaceous, buns and croissants). The possible anomalies can even be ambiguous between different products, i.e. anomalies of one product can be acceptable for another product. Thus when using a central analysis model based on artificial intelligence or on machine learning, it cannot be trained for all products. A person can train an individual model for each product. However, since the product is frequently replaced in the production line, efficient selection of the correct model is important. A schedule may be created that specifies when a change from one product to another product occurs. The method is error-prone, since it is not determined here which product is located exactly on the production line. Two approaches for solving this problem are known from the prior art. In one aspect, the manual selection of the product model can be performed by an operator. The scalability of the approach is limited and the approach is also error-prone if new products are added for the selection of possible products. On the other hand, an object recognition model can be used, which optically recognizes the product on the production line from the recorded images and loads the corresponding model. In this case, incorrect predictions from object recognition can likewise lead to the invocation of an incorrect model, so that the method is also error-prone. Disclosure of Invention The aim is to achieve an efficient selection of adapted analytical models for identifying anomalies in the object to be inspected. According to the present invention, a method is provided that allows such efficient selection of an adapted analytical model. The method according to the invention for selecting a computer-aided analysis model for use in analyzing optical anomalies of a product produced on a production line essentially comprises the steps of: Providing at least one computer-aided analysis model, Determining at least one characteristic and/or at least one production parameter of the product, -Selecting an analytical model associated with a product type from a plurality of analytical models each associated with a product type taking into account at least one determined characteristic and/or at least one determined production parameter of the product, and -Checking the product using the selected analytical model. According to the invention, additional information about the product is thus used to provide a way to integrate models from different products. The information determined about the product and/or about the production parameters allows an inference to be made about the type of product currently produced. The method according to the invention has the advantage that the products respectively produced on the production line can be reliably and effectively correlated with the adapted analytical model. It is possible to use at least one determined characteristic and/or at least one determined production parameter in the selection of the analysis model instead of or in addition to the optical analysis, for example based on the optical analysis of the image recognition software analyzing the images recorded by the camera. A production line can be understood here to be, in particular, a semiautomatic or automatic production facility, which can be equipped, for example, with a conveyor belt. On a production line, multiple or large quantities of products of the same product type can be produced in sequence in a serial operation. Within the framework of the present description, a product type is understood in particular to be a plurality of products of the same type or identical. In particular, the product type can correspond, for example, to a product model. The inspection of the product can in particular be carried out optically, for example by recording one or more images of the produ