Research on case base of panel die based on PDM

Research on the case base of panel die based on PDM

1. Introduction

in order to ensure that enterprises have sustainable innovation ability and economic benefits, how to create, accumulate and use knowledge has become the key factor for the success of current enterprises. The development of enterprises has gradually shifted from relying on capital accumulation to relying on knowledge accumulation and updating. In the field of product development, the effective use of existing knowledge is the source of product innovation and maintaining relative competitive advantage

making effective use of the empirical knowledge in the process of panel mold development can greatly improve the efficiency of mold development, but how to obtain this knowledge, especially the empirical knowledge, is a very difficult thing. Traditional knowledge management is classified according to the content of knowledge itself and stored in each sub knowledge base, ignoring the possible internal relations between different contents of knowledge. Therefore, it is of great significance to find an ideal computer expression method of design experience knowledge. So that designers can carry out mold development conveniently and flexibly, and the mold development experience can be constantly updated and supplemented with the changes of time, environment and city. The example itself contains a lot of design experience knowledge. Through the example, we can not only make full use of the design experience hidden in the example, but also more in line with human design thinking habits and engineering practice. During the development of the panel die, through the specific description of the panel and its die, and then from the case library, we can extract similar successful examples. According to the problems encountered by the die in the example and their solutions, we can predict the related problems, eliminate the potential problems, and improve the efficiency of the die development

2. Case description of panel die based on features

in the panel die product development environment, features include not only geometric/topological information, but also some non geometric attribute information, such as dimensional tolerance, roughness, material, etc. it also includes information related to the product development process, such as design intent, design version, etc., as well as information about function, performance, constraints, etc. But so far, the research on feature definition and classification has not formed a unified standard, because the definition and classification of features are restricted by the application background of feature research. Different types of product production processes are different, and the attributes of abstraction and extraction are also different. There are different opinions about which features should be included in feature modeling in the process of mold product development. According to the characteristics of process and mold development, the characteristics involved in the process of product (mold) development are summarized and divided. Panel molds are basically divided into several categories: blanking, drawing, trimming, flanging, bending, etc., and various types of molds are combined from these categories. Combined with the actual process of panel mold development, it is determined that the typical characteristic factors affecting the mold are: mold size, panel shape, panel type, mold type, panel precision, mold material, automatic device, panel material thickness and other 18 characteristics. Therefore, an example of a panel die can be described by the following quaternion

diecase={p, D, Q, o}

in the formula: P - finite non empty set, which is the feature set of the cover, including the cover category (car, truck, etc.), the cover sub category (door, hood, etc.), material, shape, accuracy, etc., and also includes the digital model of the cover, p={p1, P2... Pm},

D - non empty finite set, which is the feature information of the die, including the die category such as stamping, shaping, and the die sub category such as blanking, trimming, etc Mold materials, etc. and the digital model of the mold, d={d1, D2,... Dm}

q - set of related problems, that is, the related problems encountered in the mold development process and their solutions. This set at least includes the design description documents of the product and its mold, which is a non empty finite set, q={q1, Q2,... Qm}

o - all production processes of non empty finite sets are fully automated, which are characteristic information related to sales and quotation, o={o1, O2... Om}

where P and D feature sets are the index sets of instances

3. Organization of instance information in PDM

in the PDM system Teamcenter environment, all the data used to describe a product (such as drawings, analysis reports, test reports, process documents, etc.) are bundled together (called item in Teamcenter), so as to form a complete information model of the product. The feature information of the panel and its mold can be divided into three parts: attribute information, digital simulation and problem description and solution documents

(1) attribute information: that is, the information involved in the collection of P and D, such as panel type, mold type, etc. For the name and type information of attribute information, it can be achieved by the name of item. Other attribute information, such as panel type and material name, can be displayed by customizing the form. First, create a list of relevant values in Teamcenter and configure the POM required for the form type_ Class, and then create a new masterform class. This form class inherits from the erform class of mast's early layout and development of strategic cutting-edge materials such as superconducting materials, nano materials, graphene, etc., and writes customized information. Its code implementation is as follows:

the results are shown in Figure 1

(2) digital model of panel and mold: that is, the example describes the digital model information in feature sets P and D, including 3D digital model, 2D drawing, etc. The digital model is imported into the database as a dataset under item

(3) related problem description: that is, the features in the instance description feature set o, which are generally word documents, including problem types, problem descriptions, solutions, etc. Similar to the digital model, it is attached to item in the form of data set

combined with Teamcenter's information organization mode, the instance feature information is reasonably arranged, which is in line with the habits of product developers in the enterprise

4 panel mold instance retrieval based on similarity

4.1 calculation of similarity

judge the similarity between a mold and the existing mold by comparing the two molds in many aspects. The comparison of the degree of similarity is that the two molds are consistent in how many and what factors. Case similarity is a measure to judge the similarity between cases, which is used to find the resistance plus example when the real hydraulic oil in the case library that is most similar to the current mold passes through the system pipeline and other hydraulic components. In the actual problem solving, the attribute information of the instance is not all numerical, so it is not feasible to directly use the calculation formula to calculate the similarity. Different similarity calculation methods are used for different attribute domains. The size similarity calculation method is calculation, | L1-L2 |/l2=a

the similarity of features related to geometric dimensions is generally determined by this method. For the feature information of enumeration type, the two options need to be specified according to expert experience. For example, the similarity of drawing method is specified as follows: others, and so on, the similarity table of different attribute values of each feature is established. In similarity measurement, the importance of feature attributes is usually called weight value, which is generally difficult to judge in practice. It is usually processed through prior assumptions on the basis of empirical knowledge. The processing methods of these weight values cannot prove their correctness, which will undoubtedly add human factors to the similarity measurement of product examples and affect the accuracy of calculation results, We use the method of hierarchical weight [5] to realize the influence value of different features on the solution. Among them:

(1) the type of vehicle and parts is class I

(2) the second level is the batch size, shape complexity, maximum length, maximum width, minimum fillet and height of parts

(3) positioning requirements, accuracy, roughness, flatness, product material brand, etc. are the third level

the first level weight is 30, the second level weight is 5, and the third level weight is 1

the weight of each feature and its weight level can be adjusted in the system setting according to the actual situation to suit the actual situation of different enterprises and departments. The influence of different characteristics of the panel and its mold on the final scheme is reflected by setting the weight level to avoid the unrealistic heat conduction caused by the experimental sample and the surrounding atmosphere, which enhances the pertinence of case retrieval

4.2 instance retrieval

according to the set feature weight and its level, search the corresponding instance database, and search the similarity between the current feature attribute value and the retrieved feature attribute value. The calculation formula of the overall similarity is as follows:

, where: n - the number of all attributes that calculate the similarity

w (I) - the weight of the ith feature

e (I) - weight level of the feature

sim (U, V) - the similarity between the current feature attribute and the retrieved feature attribute

search the case base, list the retrieved cases and their solutions according to the overall similarity, and give corresponding examples and solutions for different design guidance and evaluation modules

5. Conclusion

based on the analysis of the development characteristics of the panel die, this paper analyzes the characteristics of the panel die, describes the panel die through a quaternion, and discusses the organization of the feature information in the PDM system Teamcenter. Finally, different similarity calculation methods are used for different attribute domains, and the influence of different features on the solution is realized through the method of hierarchical weight, The instance retrieval algorithm is implemented. The establishment of panel die case base can easily provide guidance for the quotation, design and manufacturing of dies in the follow-up work, and lay a foundation for the knowledge management of die enterprises. (end)