Predictive characterization model for impact cushioning curves: Configuring the predictive characterization model
- PDF / 527,893 Bytes
- 6 Pages / 612 x 792 pts (letter) Page_size
- 5 Downloads / 222 Views
9
International
Predictive Characterization Model for Impact Cushioning Curves: Configuring the Predictive Characterization Model S.-W. Lye and S. Chuchom Engineers and designers utilize mechanical properties and material behavior to assist in the design and manufacture of products. The material data obtained from standard tables tend to be general and may not correlate well with the actual material being used. To meet the design specifications, a larger number of iterative experimental tests than planned are usually conducted. This paper explores the use of neural networks as a predictive approach to characterize the impact cushioning curves so as to reduce the number of experimental tests required. Key design considerations in configuring a neural network for optimal performance are also highlighted. This approach is able to predict the points on the curves quite accurately but does have some limitations. To develop an effective predictive characterization model, the neural networks need to couple with appropriate algorithms so as to obtain a set of randomly distributed training data and generate the requisite points for curve characterization. Two algorithms are developed and found to be suitable for this purpose.
Keywords expanded polystyrene, impact properties, neural network
I
1. Introduction THE MECHANICAL PROPERTIES and behaviors of various materials can be found in engineering tables and curves. To obtain most of these data, a large number of repetitive experimental setups and tests are required. Engineers and designers use the information to assist in the design and manufacture of products and components. Nevertheless, such tables and curves have inherent limitations owing to their generality. The designed product and the test specimens may be different in terms of (a) the composition and mechanical properties of material used; (b) the varying testing parameters and conditions adopted, and (c) the manufacturing operations to produce them. For most material end users, a common industrial practice is that whenever product designers make use of engineering tables or curves, the batch of manufactured products has to be subjected to a series of iterative and laborious tests (usually the original number of tests is exceeded) in order that the final product will conform to the desired specifications. This practice is costly and can easily be improved if designers are given a more accurate and up-to-date set of material data and properties from the materials suppliers. This is easier said than done as material manufacturers and suppliers have to be concerned about their own profits and goals. Besides, the number of new and modified materials is constantly evolving. Most manufacturers would conduct their own set of tests so as to establish a set of proprietary data or curves for product differentiation and evaluation. The material characterization for product protection from impact shock is one such example. The amount of current data available on protective packaging materials such as expand-
able polystyre
Data Loading...
