Geri Dön

Otomotiv kabloları ve seçimi, denetimli öğrenme yöntemi ile kablo malzemesi seçimi

Cables used in automotive and their selection, cable material selection with supervised learning method

PDF İndir

  1. Tez No: 741507
  2. Yazar: HALE YAZGI GÜNBATI
  3. Danışmanlar: PROF. DR. ÖZCAN KALENDERLİ
  4. Tez Türü: Yüksek Lisans
  5. Konular: Elektrik ve Elektronik Mühendisliği, Electrical and Electronics Engineering
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 2022
  8. Dil: Türkçe
  9. Üniversite: İstanbul Teknik Üniversitesi
  10. Enstitü: Lisansüstü Eğitim Enstitüsü
  11. Ana Bilim Dalı: Elektrik Mühendisliği Ana Bilim Dalı
  12. Bilim Dalı: Elektrik Mühendisliği Bilim Dalı
  13. Sayfa Sayısı: 80

Özet

Bu tez giriş ve kaynak kısımlarıyla birlikte dokuz ana başlıktan oluşmaktadır. Giriş kısmında tezin amacı ve kapsamından söz edilmiştir. Otomotiv kabloları hakkında temel bilgiler verilmiş olup ağırlıkla otomotiv endüstrisinde ne tür tel kullanılır sorusu irdelenmiştir. Otomotiv kablosu seçerken hangi standartların dikkate alınacağı ve kablo çeşitleri hakkında bilgiler verilmiştir. Kablolar aslında birlikte kullanılan/kullanılması gereken parçalarıyla bir bütündür bu nedenle kablolarla birlikte kullanılan elemanlar hakkında da özellikle“otomotiv kablolama ve bağlantılarının ömrünü uzatmanın yedi yolu”başlığında yer ayrılmıştır. Doğru kablo seçimi için bir sonraki başlıkta daha detaylı bilgi verilmiş olup ardından genel kablolar hakkında bilgiler yer almıştır. Kablo tipleri, iletkenleri, yalıtkan malzemeler hakkındaki başlıklar da buranın içerisindedir. Tezde kullanılan yöntem olan makine öğrenmesi hakkında detaylı literatür taramasının ardından 7. başlıkta yalıtkan malzeme seçiminde makine öğrenmesi yöntemi uygulaması paylaşılmıştır. Sonuçlar ve Öneriler kısmının ardından kaynakların belirtilmesiyle tez sonlanmaktadır. Tezde yer alan başlıklar ana hatlarıyla ilk paragrafta bahsedilmiştir ve bu başlıklardan faydalanarak tezde incelenen durum; kablo seçiminin bir model ile yapılabilmesi. Modele kablo seçimini öğretirsek bu ne kadar güvenilir ve doğru olur? Model bunu yapmayı başarırsa, kablo seçimine ne gibi avantajlar sağlar? incelenmiştir. Bu konu incelenirken öncelikle sonucun yeterli güvenilirlikte ve doğrulukta çıkabilmesi için veri sayısının fazlalığı önemlidir. Toplanabilecek maksimum veri sayısına ulaşılmalıdır. Araştırma süresinin elverdiği ölçüde yaklaşık 300 adet veri bu konunun incelenebilmesi için toplanmıştır. Bu veriler öncelikle modele girilerek, modelin bunları öğrenmesi sağlanmıştır. Öğrenen modele artık soru sorulduğunda, verilerine dayanarak güvenilir cevabı milisaniyeler içinde verebilmektedir. Beş farklı durum denenerek model test edilmiştir ancak bu sayı arttırılabilir. Modelin cevap verdiği tüm yalıtkan malzeme sorularının yanıtları doğru çıkmıştır ve yöntemin sağlıklı çalıştığı görülmüştür. Kısa sürede yanıt veren bu sistem, kablo malzemesi seçimi yapması gereken her alanda kullanılmaya uygundur ve büyük kolaylık sağlayacaktır. Bu tezde kullanılan yöntemin doğruluk oranı yüksektir: Lojistik Regresyon'da yaklaşık %95 doğruluk, diğer algoritmalarda verilerin keskinliğinden dolayı %100 doğruluk saptanmıştır. 1 saniye içerisinde kullanıcının ihtiyacı olan bilgiyi, öğrendiği şekilde sunmaktadır. Veriler net olduğu için, makine öğrenmesi kullanımına çok uygundur ve kablo malzemesi seçiminde sadece yalıtkan seçiminde değil, endüstriden kullanıcıya geldiği noktaya kadar her aşamada bu uygulamadan faydalanılabilir. Tezde 300 veri ile yapılan bu çalışma yıllar içerisinde biriktirilen verilerin toplanarak modele öğretilmesiyle çok daha insanın karar veremeyeceği bir noktaya getirilebilir ve karmaşık durumlarda yine saniyeler içerisinde alınacak cevaplar ile uygulama tüm bu aşamalara katkı sağlar. İleriki aşamalarda modelin güvenilirliğini daha da artırmak için değişkenler artırılabilir, çıktılar talebe göre rahatlıkla düzenlenebilir. Uygulama, tezdeki haliyle kullanıcı dostu değildir, arayüzü bulunmamaktadır. Arayüz eklenerek kullanılabilir, piyasaya çıkartılabilir.

Özet (Çeviri)

This thesis consists of nine main titles together with the introduction and reference parts. In the introduction part, the aim and scope of the thesis are mentioned. Basic information about automotive cables has been given and the question of what type of wire is used in the automotive industry has been examined. Information about which standards to consider when choosing an automotive cable and cable types is given. Cables are actually a whole with the parts that are used/should be used together, so a place is reserved for the components used together with the cables, especially in the“seven ways to extend the life of automotive wiring and connections”. For the correct cable selection, more detailed information is given in the next section, followed by information about general cables. Headings about cable types, conductors, insulating materials are also included here. After a detailed literature review about machine learning, which is the method used in the thesis, the application of machine learning method in the selection of insulating materials is shared in the 7th title. After the Results and Suggestions section, the thesis ends with the references. The titles in the thesis are mentioned in the first paragraph, and the situation examined in the thesis by making use of these titles; cable selection can be made with a model. How reliable and accurate would it be if we teach the model to choose the cable? If the model manages to do this, what advantages does it give to cable selection? It has been examined. When examining this subject, first of all, it is important to have a large number of data so that the result can be obtained with sufficient reliability and accuracy. The maximum number of data that can be collected must be reached. To the extent that the research period allows, approximately 300 pieces of data were collected to examine this subject. First of all, these data were entered into the model and the model learned them. The learning model is now able to give a reliable answer within milliseconds, based on its data, when asked a question. The model has been tested by testing 5 different situations, but this number can be increased. The answers to all insulating material questions answered by the model were correct and it was seen that the method worked well. This system, which responds in milliseconds, is suitable for use in all areas where cable material selection is required and will provide great convenience. The accuracy of the method used in this thesis is high: approximately 95% accuracy in Logistic Regression, 100% accuracy in other algorithms due to the sharpness of the data. Within 1 second, it presents the information the user needs as he/she learned. Since the data is clear, it is very suitable for machine learning use and this application can be used not only in the selection of the insulation material, but also at every stage from the industry to the user. This study, which is done with 300 data in the thesis, can be brought to a point where many more people cannot decide by collecting the data accumulated over the years and teaching the model, and the application contributes to all these stages with the answers to be received within seconds in complex situations. In later stages, variables can be increased to further increase the reliability of the model, and outputs can be easily arranged according to demand. The application is not user-friendly as in the thesis, it has no interface. It can be used and released to the market by adding an interface. Logistic Regression is one of the simplest machine learning algorithms. It is easy to implement and interpret. It can be used for both binomial and polynomial cases. Logistic Regression provides a probabilistic approach to estimating the target variable. Performs well when data is linearly separable. Estimated model coefficients infer the importance of each feature. Logistic regression model training is much faster than relatively complex models. Logistic Regression gives well calibrated probabilities with classification results. This is useful for understanding the accuracy of the forecast. Logistic regression is often used as a basis for measuring performance, as it is relatively fast and easy to implement. There are also some disadvantages of logistic regression. It performs poorly when data cannot be linearly separated. More complex algorithms such as neural networks may outperform logistic regression. The presence of multiple correlations between independent variables may affect model performance. Wiring chosen for applications and maintenance of electrical systems is a critical part to ensuring the system works as designed. There are many types of automotive cable, and it can be confusing to understand which types are best suited for the application. Therefore, focusing on the wire typically used in automotive applications and discussing the two main types of insulated cable: automotive GPT primary cable and crossover cable. It also considers the importance of wire gauge size relative to the application, current draw, potential electrical resistance, and voltage drop. There are challenging situations in making automotive wiring. It must operate in adverse conditions such as extreme heat and cold, vibration, humidity, and chemicals. Chemical corrosion has increased over the years due to the use of magnesium chloride and calcium chloride instead of salt as anti-icing agents. Taken together, all these factors can cause connectors to corrode and wires to protrude, resulting in equipment downtime and repair costs. The earliest use of recognizable electrical cable was in commercialized telegraph lines, such as those stretched between Washington, D.C. and Baltimore, Maryland in 1844. These early cables were made of iron and were difficult to manufacture. Copper sulfate was used to apply a thin copper coating to improve production by lubricating the iron surface of the wires. The superconducting nature of copper was soon recognized, and copper eventually replaced these early iron conductors. In 1913, the International Electrotechnical Commission established the IACS (International Copper Standard) as a benchmark for copper's resistivity to equal 100 percent conductivity. In the 1880s, the first insulated cables were insulated with a natural latex material produced from the sap of the trees of the same name. This insulation had to be kept wet all the time or it would dry out and fail to insulate the conductors. This material has been largely replaced by rubber. By the 1890s, mass-impregnated paper insulation was being used on cables and cables at high voltages up to 10 kV. In 1906, armored cables with flexible sheath and two fabric-covered, rubber-insulated conductors were introduced. The first trials with PVC insulation were made in Germany in the 1930s, and by the end of the second world war there were significant varieties of synthetic rubber and polyethylene. In the 1950s, PVC became commercially viable and replaced rubber cables in many areas, especially in household wiring, while aluminum began to be widely used as an alternative conductor. XLPE insulation was the herald of the 1970s, replacing paper insulated cables in medium voltage applications. Automotive cables and their selection give detailed information about automotive cables in the literature summary within the cable material selection with the supervised learning method, and it exemplifies the cable selection with the machine learning method. The thesis, literature review, application of cable selection with machine learning method can be divided into two main titles. Firstly, the titles mentioned in the literature review; basic information, types used in industry, ways to extend the life of wiring and connections, right cable selection. In recent years, the need for cabling has increased due to the increasing technology and products requiring electrical connection, but since cables have different types and areas of use, it is necessary to choose the appropriate cable by considering these distinctions and criteria. Within the scope of the thesis, basic information about automotive cables, their use in industry, ways to extend their life and choosing the right cable are given. General literature about electrical cables has been explained. Based on the general literature review about electrical cables, it was desired to make the cable selection an application. The machine learning method was mentioned, and electrical cable selection was tried with this method. Cable selection was made to the machine in five different scenarios and the results were evaluated.

Benzer Tezler

  1. Tez No

    129309

    Commodity management supplier (CMS) selection at Ford Motor Company by utilizing logistic regression model

    Ford Motor Company'de ürün grubu yöneticisi (ÜGY) tedarikçilerinin lojistik regresyon kullanılarak seçimi

    SABRİYE SİBEL ERKILIÇ

    Yüksek Lisans

    İngilizce

    İngilizce

    2002

    Endüstri ve Endüstri MühendisliğiBoğaziçi Üniversitesi

    Endüstri Mühendisliği Ana Bilim Dalı

    PROF. DR. İLHAN OR

  2. Tez No

    887688

    Nonlinear model predictive control with real time iteration for F-16 attitude control

    F-16 duruş kontrolü için gerçek zamanlı iterasyon iledoğrusal olmayan model öngörülü kontrol

    SİYAMİ GÜRKAN KUZUCU

    Yüksek Lisans

    İngilizce

    İngilizce

    2024

    Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrolİstanbul Teknik Üniversitesi

    Kontrol ve Otomasyon Mühendisliği Ana Bilim Dalı

    PROF. DR. MUSTAFA DOĞAN

  3. Tez No

    152251

    Bir dizel motorunun performans testi için aşamalı istatistiksel motor haritalama modellerinin geliştirilmesi

    The development of two stage statistical engine mapping models for performance test of a diesel engine

    EMRE ESENTÜRK

    Yüksek Lisans

    Türkçe

    Türkçe

    2004

    Makine Mühendisliğiİstanbul Teknik Üniversitesi

    Makine Mühendisliği Ana Bilim Dalı

    PROF. DR. CAN ÖZSOY

  4. Tez No

    558694

    Seat backrest vibration improvement of a commercial vehicle

    Bir ticari araç koltuğunun sırt titreşimlerinin iyileştirilmesi

    AHMET AVŞAR

    Yüksek Lisans

    İngilizce

    İngilizce

    2019

    Mühendislik Bilimleriİstanbul Teknik Üniversitesi

    Makine Mühendisliği Ana Bilim Dalı

    PROF. DR. HALUK EROL

  5. Tez No

    639983

    Borsa İstanbul'da işlem gören savunma sanayi şirketlerinin finansal performans analizi

    Financial performance analysis of defense industry companies in Borsa İstanbul

    KERIM NURYYEV

    Yüksek Lisans

    Türkçe

    Türkçe

    2020

    İşletmeAfyon Kocatepe Üniversitesi

    İşletme Ana Bilim Dalı

    DR. ÖĞR. ÜYESİ SERDAR ÖGEL

Geri Dön