Wyniki wyszukiwania - Biblioteka Nauki

Ten serwis zostanie wyłączony 2025-02-11.

Nowa wersja platformy, zawierająca wyłącznie zasoby pełnotekstowe, jest już dostępna.
Przejdź na https://bibliotekanauki.pl

Ograniczanie wyników

2 Archives of Foundry Engineering

2 2008

Znaleziono wyników: 2

Liczba wyników na stronie

Wyniki wyszukiwania

Wyszukiwano:
w słowach kluczowych: knowledge rules

Sortuj według:

Ogranicz wyniki do:

Comparison of selected tools for generation of knowledge for foundry production

100%

Perzyk M. , Soroczyński A.

tom Vol. 8, iss. 4

163-166

Two types of data mining tools, suitable for semi-automatic generation of knowledge in a form of logic rules, are presented in the paper: decision (classification) trees and rough sets theory algorithms. A comparative evaluation of rules obtained by these two methods, used for decision concerning application of feeders for grey iron castings, is performed. Data sets obtained as readouts form a semi-empirical nomograph of Holzmüller and Wlodawer were used for the testing. It was found that both methods lead to similar rules, which are also in agreement with the foundry practice. However, the decision trees were unable to provide some important and reliable rules, which were generated by the rough sets theory algorithm and they can also generate rules which are not supported by the training data.

Possibilities of decision trees applications for improvement of quality and economics of foundry production

80%

Perzyk M. , Soroczyński A. , Biernacki R.

tom Vol. 8, iss. 1 spec.

261--268

The key activity areas, related to quality and economics of foundry production, are presented: designing of manufacturing processes, control of production processes as well as analysis of root causes of process faults and irregularities, are presented. Possibilities of utilization of data mining methods, including decision (classification) trees type learning systems, are indicated. In particular, the role of that kind of tools in decision making concerning selection of process type and optimum materials and parameters as well as in identification of process excessive variations, similarly like with the control charts, are discussed. Evaluation results of classification systems form the viewpoint of their applicability, accuracy and software availability are presented, including decision trees, naïve Bayesian classifier, rough sets theory, direct rule induction methods as well as artificial neural networks. In the final part of the paper a knowledge in the form rules obtained form classification trees is demonstrated using the example of decision making related to application of risers for grey cast iron castings.