Ağır Vasıta Hava Kompresörü Arıza Durumlarının Naive Bayes Sınıflandırıcısı Kullanılarak Analizi
No Thumbnail Available
Date
2021
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Ejosat
Open Access Color
GOLD
Green Open Access
No
OpenAIRE Downloads
OpenAIRE Views
Publicly Funded
No
Abstract
Hava kompresörleri ağır vasıta fren sistemlerinin hava ile beslenmesi için hayati öneme sahiptir. İçerisinde bulundurduğu piston biyel mekanizması ile havanın istenilen basınca ulaştırılarak (10-12.5 bar) tanka depo edilmesini sağlamaktadır. Planlı bakımlarda bileşenlerinin cüzi ücretlerle değişimleri yapılabilirken, arıza durumlarında aracın yolda kalmasına sebep olabilmekte ve yüksek miktarda ücret ve plansız zaman kaybı ile onarımı yapılabilmektedir. Beklenmedik arıza süreleri ağır vasıtaların taşıdıkları ürünleri müşteriye geç teslim edilmesine sebep olabilmekte ve müşteri memnuniyetsizliğine sebep olmaktadır. Bu çalışma ile hava kompresöründe gerçekleşebilecek arıza durumları araştırılmış ve firma Ar-Ge biriminde oluşturulan test düzeneği ile belirlenen arızalar manuel olarak gerçekleştirilmiştir. Elde edilen veriler Dewesoft yazılımı ile kayıt altına alınmıştır. Python yazılımı kullanılarak Naive Bayes Sınıflandırıcısı modelleri oluşturulmuştur. Toplam 23.987 veri kullanılmıştır. Bu verilerin %80’lik kısmı ile modeller eğitilmiş ve %20 ‘lik kısmı ile arıza tahmininde bulunulmuştur. Sırasıyla eğitim ve test verisi hatalı tahmin sayıları; Bernoulli Naive Bayes Sınıflandırıcısı için 18062 – 4550, Multinominal Naive Bayes Sınıflandırıcısı için 16654 –4210, Gaussian Naive Bayes Sınıflandırıcısı için 961–258 veri hatalı tahminde bulunmuştur. Gaussian Naive Bayes sınıflandırıcısının bariz farkla doğru kararlar verdiği gözlemlenmiştir. Ayrıca sınıflandırma metrikleri incelenerek elde edilen sonuçlar değerlendirilmiştir.
Air compressors are vital for the air supply of heavy vehicle braking systems. With the piston connecting rod mechanism it has in it, it ensures that the air is delivered to the desired pressure (10-12.5 bar) and stored in the tank. While changes of its components can be made at small fees during planned maintenance, it can cause the vehicle to stay on the road in case of failure and can be repaired with a high amount of fees and unplanned loss of time. Unexpected failure times can cause heavy vehicles to deliver their products to the customer late and cause customer dissatisfaction. In this study, the fault situations that may occur in the air compressor were investigated and the deciencies determined by the test device created in the company's R &D unit were performed manually. The obtained data were recorded with Dewesoft software. Naive Bayes Classifier models were created using Python software. A total of 23,987 data were used. Models were trained with 80% of the total data and fault prediction was made with 20%. The training and test data, respectively, are incorrect prediction numbers; 18062–4550 for Bernoulli Naive Bayes Classifier, 16654-4210 for Multinomial Naive Bayes Classifier, 961-258 for Gaussian Naive Bayes Classifier data has been incorrectly estimated. It has been observed that the Gaussian Naive Bayes classifier makes correct decisions with obvious difference. In addition, the results obtained by examining the classification metrics were evaluated.
Air compressors are vital for the air supply of heavy vehicle braking systems. With the piston connecting rod mechanism it has in it, it ensures that the air is delivered to the desired pressure (10-12.5 bar) and stored in the tank. While changes of its components can be made at small fees during planned maintenance, it can cause the vehicle to stay on the road in case of failure and can be repaired with a high amount of fees and unplanned loss of time. Unexpected failure times can cause heavy vehicles to deliver their products to the customer late and cause customer dissatisfaction. In this study, the fault situations that may occur in the air compressor were investigated and the deciencies determined by the test device created in the company's R &D unit were performed manually. The obtained data were recorded with Dewesoft software. Naive Bayes Classifier models were created using Python software. A total of 23,987 data were used. Models were trained with 80% of the total data and fault prediction was made with 20%. The training and test data, respectively, are incorrect prediction numbers; 18062–4550 for Bernoulli Naive Bayes Classifier, 16654-4210 for Multinomial Naive Bayes Classifier, 961-258 for Gaussian Naive Bayes Classifier data has been incorrectly estimated. It has been observed that the Gaussian Naive Bayes classifier makes correct decisions with obvious difference. In addition, the results obtained by examining the classification metrics were evaluated.
Description
ORCID
Keywords
Ağır vasıta hava kompresörü arızaları, Bernoulli – Multinominal - Gaussian Naive Bayes sınıflandırıcısı, Makine öğrenmesi, Heavy vehicle air compressor fault, Bernoulli – Multinominal - Gaussian Naive Bayes classifier, Machine learning, Engineering, Mühendislik, Heavy Vehicle Air Compressor Fault;Bernoulli – Multinominal - Gaussian Naive Bayes Classifier;Machine Learning., Ağır Vasıta Hava Kompresörü Arızaları;Bernoulli – Multinominal - Gaussian Naive Bayes Sınıflandırıcısı;Makine Öğrenmesi
Turkish CoHE Thesis Center URL
Fields of Science
0202 electrical engineering, electronic engineering, information engineering, 02 engineering and technology
Citation
WoS Q
N/A
Scopus Q
N/A

OpenCitations Citation Count
1
Source
European Journal of Science and Technology
Volume
31
Issue
1
Start Page
796
End Page
800
PlumX Metrics
Citations
CrossRef : 1
Captures
Mendeley Readers : 3
Downloads
1
checked on Feb 03, 2026
Google Scholar™


