Farklı lityum iyon piller için batarya şarj durumu tahmini
Date
2024-07-31
Authors
Tekin, Merve
Journal Title
Journal ISSN
Volume Title
Publisher
Bursa Uludağ Üniversitesi
Abstract
Hibrit ve elektrikli araçlar taşımacılık sektörünün iklim değişikliğine olan etkilerini azaltmada en ümit verici teknolojilerdir. Ancak elektrikli araçların anahtar bileşeni olan lityum-iyon bataryalarla ilgili geliştirilmesi gereken hususlar mevcuttur. Lityum-iyon bataryaların güvenli ve verimli bir şekilde çalışması için bir Batarya Yönetim Sistemi (BYS) tarafından kontrol edilmesi gerekir. BYS akım, voltaj, sıcaklık, batarya şarj durumu (BŞD), batarya yaşlanma durumu (BYD) ve batarya güç durumu (BGD) gibi batarya iç durumlarını sürekli olarak kontrol ederek bataryayı aşırı şarj/deşarja, hücreler arasındaki dengesizliklere ve termal kaçaklara karşı korur. BŞD bataryanın anlık durumunu kontrol etmede kritiktir. Ayrıca diğer batarya durumları ile de ilişkili olduğundan BŞD’nin doğru tahmin edilmesi BYS’nin etkili ve verimli bir şekilde çalışmasındaki anahtar faktörlerden biridir. Bu çalışmada hibrit ve elektrikli araç bataryalarını domine eden Nikel Mangan Kobalt (NMC), Nikel Kobalt Alüminyum (NCA) ve Lityum Demir Fosfat (LFP) kimyalarına sahip piller için farklı tahmin yöntemleri karşılaştırılmıştır. Kullanılan yöntemler modellemeye dayalı yaklaşımlar olan Genişletilmiş Kalman Filtresi (GKF) ve Kokusuz Kalman Fitresi (KKF) ve bilgisayar öğrenmesine dayalı Kapı Özyinelemeli Birimler (GRU)’dir. Geliştirilen tahmin algoritmalarının aracın sürüş esnasındaki yüksek dinamik davranışı altında dahi doğru tahminler gerçekleştirmesi batarya kontrolü, enerji yönetimi ve kullanıcının doğru bilgilendirilmesi bakımından önemlidir. Bu nedenle kullanılan yöntemlerin BŞD tahmin performansları bir aracın sürüş davranışını en gerçekçi şekilde yansıtan ve en güncel sürüş çevrimi olan WLTP için karşılaştırılmıştır. Çalışmanın sonucunda modellemeye dayalı yöntemlerle daha düşük hesaplama maliyeti ve daha iyi tahmin performansı elde edilmiştir. Üç pil kimyası için de en iyi sonuçlar KKF ile sağlanmıştır.
Hybrid and electric vehicles are the most promising options for reducing the transportation sector's impact on climate change. However, some concerns need to be addressed with the lithium-ion battery pack, which is an essential component of electric vehicles. Lithium-ion batteries require a Battery Management System (BMS) to operate safely and efficiently. The BMS protects the battery against overcharging and discharging, cell imbalances, and thermal runaway by continuously tracking internal battery states such as current, voltage, temperature, battery state of charge (SoC), state of health (SoH), and state of power (SoP). As the SoC is vital in controlling the battery's instantaneous state and is also correlated with other battery states, accurate SoC estimate is one of the most important components in the BMS's effective and efficient operation. This study compares different estimation approaches for battery cells using Nickel Manganese Cobalt (NMC), Nickel Cobalt Aluminum (NCA), and Lithium Iron Phosphate (LFP) chemistries, which dominate hybrid and electric vehicle batteries. Modeling-based approaches such as the Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF) are used, as well as machine learning-based Gated Recurrent Unit (GRU). It is critical that the developed estimation algorithms are compatible with the vehicle's ever-changing, dynamic behavior while driving and provide accurate predictions in terms of battery control, energy management, and user information. For this reason, the SoC estimation performances of the approaches used were compared using the WLTP (Worldwide Harmonized Light Vehicles Test Procedure) driving cycle, which is the most recent and realistic representation of a vehicle's driving behavior. As a result of the study, lower computational cost and better prediction performance were obtained with modeling-based methods. The best results for all three battery chemistries were obtained with UKF.
Hybrid and electric vehicles are the most promising options for reducing the transportation sector's impact on climate change. However, some concerns need to be addressed with the lithium-ion battery pack, which is an essential component of electric vehicles. Lithium-ion batteries require a Battery Management System (BMS) to operate safely and efficiently. The BMS protects the battery against overcharging and discharging, cell imbalances, and thermal runaway by continuously tracking internal battery states such as current, voltage, temperature, battery state of charge (SoC), state of health (SoH), and state of power (SoP). As the SoC is vital in controlling the battery's instantaneous state and is also correlated with other battery states, accurate SoC estimate is one of the most important components in the BMS's effective and efficient operation. This study compares different estimation approaches for battery cells using Nickel Manganese Cobalt (NMC), Nickel Cobalt Aluminum (NCA), and Lithium Iron Phosphate (LFP) chemistries, which dominate hybrid and electric vehicle batteries. Modeling-based approaches such as the Extended Kalman Filter (EKF) and Unscented Kalman Filter (UKF) are used, as well as machine learning-based Gated Recurrent Unit (GRU). It is critical that the developed estimation algorithms are compatible with the vehicle's ever-changing, dynamic behavior while driving and provide accurate predictions in terms of battery control, energy management, and user information. For this reason, the SoC estimation performances of the approaches used were compared using the WLTP (Worldwide Harmonized Light Vehicles Test Procedure) driving cycle, which is the most recent and realistic representation of a vehicle's driving behavior. As a result of the study, lower computational cost and better prediction performance were obtained with modeling-based methods. The best results for all three battery chemistries were obtained with UKF.
Description
Keywords
Lityum-iyon bataryalar, Elektrikli araçlar, Batarya yönetim sistemi, Şarj durumu tahmini, Kalman filtresi, Yapay sinir ağları, Lithium-ion batteries, Elektric vehicles, Battery management system, State of charge estimation, Kalman filter, Artificial Neural networks