<abstract xmlns="http://www.w3.org/1999/xhtml"><p>An attempt of designing artificial neural networks for empirical laboratory test results tracers No. 5, No. 7 and No. 8 was introduced in the article. These tracers are applied in cartridges with calibres from 37 mm to 122 mm which are still used and stored both in the marine climate and land. The results of laboratory tests of tracers in the field of over 40 years of tests have been analysed. They have been properly prepared in accordance with the requirements that are necessary to design of neural networks. Only the evaluation module of these tracers was evaluated, because this element of tests, fulfilled the necessary assumptions needed to build artificial neural networks. Several hundred artificial neural networks have been built for each type of analysed tracers. After an in-depth analysis of received results, it was chosen one the best neural network, the main parameters of which were described and discussed in the article. Received results of working built of neural networks were compared with previously functioning manual evaluation module of these tracers. On the basis conducted analyses, proposed the modification of functioning test methodology by replacing the previous manual evaluation modules through elaborated automatic models of artificial neural networks. Artificial neural networks have a very important feature, namely they are used in the prediction of specific output data. This feature successfully used in diagnostic tests of other elements of ammunition.</p></abstract>

An attempt of designing artificial neural networks for empirical laboratory test results tracers No. 5, No. 7 and No. 8 was introduced in the article. These tracers are applied in cartridges with calibres from 37 mm to 122 mm which are still used and stored both in the marine climate and land. The results of laboratory tests of tracers in the field of over 40 years of tests have been analysed. They have been properly prepared in accordance with the requirements that are necessary to design of neural networks. Only the evaluation module of these tracers was evaluated, because this element of tests, fulfilled the necessary assumptions needed to build artificial neural networks. Several hundred artificial neural networks have been built for each type of analysed tracers. After an in-depth analysis of received results, it was chosen one the best neural network, the main parameters of which were described and discussed in the article. Received results of working built of neural networks were compared with previously functioning manual evaluation module of these tracers. On the basis conducted analyses, proposed the modification of functioning test methodology by replacing the previous manual evaluation modules through elaborated automatic models of artificial neural networks. Artificial neural networks have a very important feature, namely they are used in the prediction of specific output data. This feature successfully used in diagnostic tests of other elements of ammunition.

Applying of Neural Networks for Testing of Tracers with using of Empirical Data

Military Institute of Armament Technology

Maritime Technical Journal

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

{"article-title":"Applying of Neural Networks for Testing of Tracers with using of Empirical Data"}

An attempt of designing artificial neural networks for empirical laboratory test results tracers No. 5, No. 7 and No. 8 was introduced in the article. These...

Applying of Neural Networks for Testing of Tracers with using of Empirical Data

Pubblicato online: 30 set 2019

Pagine: 5 - 23

Ricevuto: 18 feb 2019

Accettato: 06 set 2019

DOI: https://doi.org/10.2478/sjpna-2019-0015

Parole chiave
neural networks, tracer, laboratory tests, importance classes, evaluation module, hidden layer, neuron

© 2019 Dariusz Ampuła, published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Applying of Neural Networks for Testing of Tracers with using of Empirical Data

Pubblicato online: 30 set 2019

Pagine: 5 - 23

Ricevuto: 18 feb 2019

Accettato: 06 set 2019

DOI: https://doi.org/10.2478/sjpna-2019-0015

Parole chiaveneural networks, tracer, laboratory tests, importance classes, evaluation module, hidden layer, neuron

© 2019 Dariusz Ampuła, published by Sciendo

This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 3.0 License.

Parole chiave
neural networks, tracer, laboratory tests, importance classes, evaluation module, hidden layer, neuron