Publication: İki serbestlik dereceli araç simülatörü analizi ve uygulaması
Abstract
Günümüzde simülatörler ve simülasyon teknikleri birçok alanda kullanılmaktadır. Ülkemizde ve dünyada bu tekniğin önemi giderek artmaktadır. Simülasyonlar, problemler hakkında objektif bilgiye sahip olmamızı sağlar. Geçici ve kalıcı rejimlerin sistem dinamiği üzerinden gözlemlenmesine yardımcı olurlar. Bu çalışmada 2 DoF araç simülatörü braketleri için sürüş senaryoları ile desteklenmiş topoloji optimizasyonu sunulmuştur. İlk olarak 14 DoF a sahip bir binek model referans uygulaması Simulink’te belirlenmiştir. Ardından aracın dinamik performansını analiz etmek için yaygın olarak kullanılan sürüş senaryoları (Constant Radius, Double Lane Change, Fishhook, Increasing Steer, Sine with Dwell ve Swept Sine) 14 DoF araç modelinde koşturulmuştur. Elde edilen analiz sonuçlarında, xyz eksenlerinde oluşan ivmeler kaydedilmiştir. Burada oluşan ivmeler için minimum ve maksimum ivme değerleri her sürüş senaryosu için ayrı ayrı gruplandırılmıştır. SOLIDWORKS Simulation ortamında sürüş senaryolarından elde edilen ivmeler 2 DoF araç simülatörü üzerinde koşturulup stress deformasyon analizi yapılmıştır. Stress deformasyon analizleri esnasında 2 DoF araç simülatörü üzerinde oluşan reaksiyon kuvvetleri doğrultusunda lineer aktüatör ve eksen kuvvetleri belirlenmiştir. Belirlenen eksen kuvvetleri altında braketler topoloji optimizasyonuna tabi tutulmuştur. Elde edilen yenilikçi tasarım rötuşlanarak üretime uygun hale getirilmiştir. Şekil optimizasyonu yapılan braketler 2 DoF araç simülatörü üzerinde sürüş senaryolarından elde edilen ivme değerleri ile tekrar koşturulup stress ve deformasyon analizi yapılmıştır. İlgili bilgiler ışığında tasarlanan sistem imal edilerek montajı gerçekleştirilmiştir. Ardından sistem uygulaması yapılmıştır. Uygulama ve kontrol kısmında, IoT ve Sinir Ağı tabanlı 2 DoF araç simülatörü için sürüş senaryolarının dağılımı sunulmuştur. 2 DoF manipülatörde 3 eksende uygun ivmelerin aktarılması lineer olmayan bir problemi çağrıştırdığından kontrolör yapısı Yapay Sinir Ağı tabanlı kontrolör olarak seçilmiştir. Yapay Sinir Ağı hesaplamalarını gerçekleştirmek için araç simülatöründe kullanılan mikro denetleyicinin sınırlı işlem kapasitesi ve hızı nedeniyle IoT tabanlı bilgi işlem ve veri aktarımı seçilmiştir. İlk olarak, araç simülatörünü tanımlamak ve sinir ağı için eğitim verilerini oluşturmak için bir açık döngü ölçümü gerçekleştirilir. Daha sonra eksenlerdeki hızlanma verileri ve kontrol sinyalleri kaydedilir. İkinci olarak, günlüğe kaydedilen verilerle sinir ağı eğitimi gerçekleştirilir. Son olarak eğitilen sinir ağı çeşitli sürüş manevraları ile test edilerek çalışma tamamlanmıştır.
Today, simulators and simulation techniques are used in many fields. The importance of this technique is increasing in our country and in the world. Simulations allow us to have objective information about problems. They help to observe temporary and permanent regimes through system dynamics. In this study, topology optimization supported by driving scenarios for 2 DoF vehicle simulator brackets is presented. First, a passenger model reference application with 14 DoF was identified in Simulink. Then, commonly used driving scenarios (Constant Radius, Double Lane Change, Fishhook, Increasing Steer, Sine with Dwell and Swept Sine) were run on 14 DoF vehicle models to analyze the dynamic performance of the vehicle. In the analysis results obtained, the accelerations in the xyz axes were recorded. The minimum and maximum acceleration values for the accelerations formed here are grouped separately for each driving scenario. In the SOLIDWORKS Simulation environment, the accelerations obtained from the driving scenarios were run on 2 DoF vehicle simulators and stress deformation analysis was performed. During the stress deformation analysis, linear actuator and axis forces were determined in line with the reaction forces on the 2 DoF vehicle simulator. Under the determined axial forces, the brackets were subjected to topology optimization. The resulting innovative design has been retouched and made suitable for production. Shape-optimized brackets were rerun on 2 DoF vehicle simulators with acceleration values obtained from driving scenarios, and stress and deformation analysis were performed. The system designed in the light of the relevant information was manufactured and assembled. Then the system application was made. In the application and control part, the distribution of driving scenarios for 2 DoF vehicle simulators based on IoT and Neural Network is presented. Since the transfer of appropriate accelerations in 3 axes in 2 DoF manipulators evokes a non-linear problem, the controller structure was chosen as an Artificial Neural Network based controller. Due to the limited processing capacity and speed of the microcontroller used in the vehicle simulator to perform the Artificial Neural Network calculations, IoT-based computing and data transfer were chosen. First, an open loop measurement is performed to define the vehicle simulator and generate the training data for the neural network. The acceleration data and control signals on the axes are then recorded. Second, neural network training is performed with the logged data. Finally, the trained neural network was tested with various driving maneuvers and the study was completed.
Today, simulators and simulation techniques are used in many fields. The importance of this technique is increasing in our country and in the world. Simulations allow us to have objective information about problems. They help to observe temporary and permanent regimes through system dynamics. In this study, topology optimization supported by driving scenarios for 2 DoF vehicle simulator brackets is presented. First, a passenger model reference application with 14 DoF was identified in Simulink. Then, commonly used driving scenarios (Constant Radius, Double Lane Change, Fishhook, Increasing Steer, Sine with Dwell and Swept Sine) were run on 14 DoF vehicle models to analyze the dynamic performance of the vehicle. In the analysis results obtained, the accelerations in the xyz axes were recorded. The minimum and maximum acceleration values for the accelerations formed here are grouped separately for each driving scenario. In the SOLIDWORKS Simulation environment, the accelerations obtained from the driving scenarios were run on 2 DoF vehicle simulators and stress deformation analysis was performed. During the stress deformation analysis, linear actuator and axis forces were determined in line with the reaction forces on the 2 DoF vehicle simulator. Under the determined axial forces, the brackets were subjected to topology optimization. The resulting innovative design has been retouched and made suitable for production. Shape-optimized brackets were rerun on 2 DoF vehicle simulators with acceleration values obtained from driving scenarios, and stress and deformation analysis were performed. The system designed in the light of the relevant information was manufactured and assembled. Then the system application was made. In the application and control part, the distribution of driving scenarios for 2 DoF vehicle simulators based on IoT and Neural Network is presented. Since the transfer of appropriate accelerations in 3 axes in 2 DoF manipulators evokes a non-linear problem, the controller structure was chosen as an Artificial Neural Network based controller. Due to the limited processing capacity and speed of the microcontroller used in the vehicle simulator to perform the Artificial Neural Network calculations, IoT-based computing and data transfer were chosen. First, an open loop measurement is performed to define the vehicle simulator and generate the training data for the neural network. The acceleration data and control signals on the axes are then recorded. Second, neural network training is performed with the logged data. Finally, the trained neural network was tested with various driving maneuvers and the study was completed.