Вы находитесь на странице: 1из 4

New Die Design System for Injection Molding using Multivariate Analysis

Shinnosuke NITOH
Computer Science, Graduate School, College of Engineering, Nihon University

Yoshikazu KOBAYASHI and Kenji SHIRAI


College of Engineering, Nihon University

Abstract
Past injection mold shape data and processing data are often used in injection mold design. In these cases, the building of a database on past examples and retrieval of similar data shorten design time. The evaluation method multivariate analysis is used to calculate the similarity level between data input by the designer and data in the databases related to the injection mold composition parts that the designer intends to build and past design data. In this system, the designer edits corrections, changes, etc. using a list of injection mold composition parts and list of shape features, which have been retrieved as the most similar by calculating the similarity level, and a 3D model of the injection mold, which can be displayed and processed by DirectX on the web.Even users with only a slight knowledge of injection mold design can design injection molds efficiently using this system, and at design times about 50% shorter than conventional methods.

1. Introduction
In recent years, injection molds are designed using the three-dimension solid system1)2). Generally past design data is used in most injection mold design jobs3). However, it is difficult to search for data stored on computer quickly. For such knowledge succession problems in the manufacturing industry, the formation process of knowledge is theorized or formulized as in artificial engineering4)and TRIZ . Based on this, attempts are being made to construct mechanisms which support engineers. With the system introduced in this report, first parts composing injection molds designed in the past or processed parts were registered6). Next, the data of parts composing injection molds was compared. In the comparison, the degree of similarity such as types of mold parts, part size, and number of parts, etc. were taken as parameters. Since there were multiple parameters, we prototyped a computer-aided design system for injection molds which can retrieve data by multi-variate analysis7).
5)

2. System Configuration
The system proposed in this study mainly consists of three subsystems; an injection mold information input unit as shown in Figure 1, a similarity evaluation unit, and a knowledge base unit. The knowledge base unit stores information on the components of injection molds, features of injection mold structures, and injection mold 3D model data. The design information input unit inputs parameters on basic parts, additional features, etc. composing the injection mold designed using dialog. The similarity evaluation unit on the other hand evaluates similarities between input data and the data of tables Figure 1: System configuration in the knowledge base unit. The knowledge base unit was classified into 111 kinds of form features, including 180 kinds of parts, such as guide pin constituting an injection mold, ejector rod, sprue bush, hole and slot machining parts, special parts, and unit mechanism. The main injection mold composition data stored in the knowledge base unit are data on additional machining, inclusion of parts, injection mold features, design features, mold-base, mechanism data, standard mold-base specification changes, etc. There are also integrated data consolidating these seven data. Retrieval results are displayed in dialog form as shown in Figure 2. 3D models of retrieval results are displayed in four parts. Displays include list-displays of components of molds searched, inventory information on parts, machining information, part price, part name of selection parts, machining time, prices, total machining time and total prices, and displays of 3D models of retrieval results. In the 3D models displayed, parts are differentiated by color. Designers are able to change to any viewpoint using the mouse.

M_12 M_05 M_06 M_10 M_04 M_09 M_08 Image model M_03 M_01 M_01 M_11

M_07 M_02

Figure 2:

Display by 3D model of result and mold parts information

Figure 3: Display by Euclid space of an analysis result

3. Similarity Evaluation
This system was designed to evaluate similarities between composition data of injection molds designed in the past and composition data of injection molds visualized by the designer and search for data. Similarities are evaluated by quantification of multi-variate analysis and analysis of main components which are statistical techniques. Quantification is the method of spatially classifying qualitative variables while the analysis of main components is the method of spatially classifying quantitative variables. Since similar molds can be designed using past injection mold composition data the knowledge of experts play an important role in such past design data, the knowledge of which is composed of certain patterns. Moreover, injection molds are composed by combining form features. Components and injection mold structures are therefore similar. But there exist no two same molds. As parts which constitute an injection mold can be treated as qualitative data, the Euclid distance from the image data is calculated by the quantification method to evaluate similarity. The results are displayed on the Euclid space as shown in Figure 3. Because the numerical data of injection mold components are quantitative data, similarity evaluation of component size, quality, and attached position, etc. is performed using the analysis of main components.

4. Comparison of Analysis Results 4.1 Comparison of Mold Component Plate and Size
An example of analysis results by this system are described here. The input data of the analysis result example are parameters on injection mold features, hole and slot machining, types of parts, and unit mechanism. The output results are the Model of the overall image of the mold as shown in Figure 4 and hole and slot machining, and injection mold components as shown in Figures 5, 6, 7, 8, 9, and 10. Model A is the Model visualized by the designer. Model B is the result of retrieval of similar injection molds using all four parameters which injection mold features, hole and slot machining, types of parts, and unit mechanism as the input data. Model C is the result of retrieval of similar injection molds using injection mold feature as the input data.

Figure 4: Comparison with the image molding type and the result of a analysis (Molding composition and dimension)

Figure 5:

Designers image of molding

(Additional machining and installation parts (Model A))

Figure 6: The comparison of the additional machining characteristics as a result of an analysis (Model B)

Figure 7: The comparison of the additional machining characteristics as a result of an analysis (Model C)

Figure 8: The comparison of the additional machining characteristics as a result of an analysis (Model D)

Figure 9: The comparison of the additional machining characteristics as a result of an analysis (Model E)

Figure 10: The comparison of the additional machining characteristics as a result of an analysis (Model F)

Model D is the result of retrieval of similar injection molds using hole and slot machining as the input data. Model E is the result of retrieval of similar injection molds using types of parts as the input data. Model F is the result of retrieval of similar injection molds using unit mechanism as the input data. If each Model is compared visually, Model C is not similar while Models B, D, E, and F are. Next, the plate size of each Model is compared. Models C and F are completely different plates while Models B, D, and E are comparatively of the same size. This is based on the difference in built-in parts and internal structure.

4.2 Comparison of Form Feature and Built-in Parts


Next, the hole and slot machining and types of parts of each Model are compared. Figure 6 shows the hole and slot machining, and types of parts visualized by the designer. There are seven types of hole and slot machining and ten types of parts. In the example of Model B, there are five types of the same hole and slot machining, and eight types of parts. The same comparison was performed for Models C, D, E, and F. In the example of Model C, there are two kinds of the same hole and slot machining, and no types of part were the same. In the example of Model D, there were six kinds of the same hole and slot machining, and seven types of parts. In the example of Model E, there were five kinds of the same hole and slot machining, and nine types of parts. In the example of Model F, there were three kinds of the same hole and slot machining, and two types of parts. These results indicate that importance given to mold structure by input data which injection mold features, hole and slot machining, type of part, and unit mechanism. Consequently, injection mold size and hole and slot machining are correlated with built-in parts and internal structure, and the size of the whole injection mold and the kind of hole and slot machining can be determined from built-in part and internal structure. However, although Models D and E must search for data with all the same elements as the injection mold components visualized by the designer, different data is searched in some cases. This is considered to be search of data containing some wrong information, where a part of a mold designed in the past was processed at an unintended area. For Models C and F, since the data is about the internal structure of an injection mold component, hole and slot machining, and parts corresponding to the mechanism were searched.

5. Conclusion
An injection mold design system applying multi-variate analysis was prototyped, and the following conclusions were obtained. (1) Since retrieval reflecting data composing injection molds designed in the past can be performed, even users with little experience in injection mold design can design easily. Moreover, design time is about half that of conventional methods. (2) Injection mold components such as components, design data, etc. of past injection molds, are managed into seven parts, and the data which the designer needs from this point can be searched quickly. In future studies, the authors plan to construct knowledge base systems added with functions which can perform searches taking into consideration special parts composing the mold and special structures or knowledge base systems taking into consideration prioritization based on designer instructions, as well as to investigate retrieval methods. References
1)O.J Canciglieri, R.I.M. Young: A Multi-Viewpoint Reasoning System in Design for Injection Moulding, 1997, Conference Proceedings of International Conference and Exhibition on Design and Production of Dies and Molds, Molding, p21-26 2)J. P. Kruth, R. Willems, D. Lecluse: A Design System using High Level Mould Objects, 1997, Conference Proceedings of International Conference and Exhibition on Design and Production of Dies and Molds, Molding, p39-44 3) Shuji KUBO, Masayuki ASAI, Yoshiro KASHIWABARA, Kauai TSHUKAHARA: Construction of the BPR concept of a product factory, 1999, Die & Mould Technology, vol14, no.7. 4) Toshiharu TAURA: Knowledge in engineering, artificial thing engineering, 1995.4, the Japan Society for Precision Engineering, p61. 5) Victor R. Fey, Eugene, I. Riving, Yoder HATAYAMA: 1997, A guide to TRIZ, Nikon Kogyo Shim bun Ltd. 6) Shinnosuke NITOH, Yoshikazu KOBAYASHI and Kenji SHIRAI: Development of the form similarity evaluation system 2001, and the injection mold design, the collection of the Japan Society for Precision Engineering spring convention academic lecture meeting treatises in the 2001 fiscal year, p147. 7) Koichi SUGIYAMA: A guide to multi- variable data analysis, 1983.4, Asakura Publishing Co., Ltd.

Вам также может понравиться