Yazar "Ayana, Omer" seçeneğine göre listele

Listeleniyor 1 - 2 / 2
  • [ X ]
    Öğe
    Industrial Fault Detection and Classification with the Optimal Windows Size Approach
    (IEEE, 2024) Ayana, Omer; Inan, Ali
    Various important issues in industrial production processes such as product quality, process safety and supply continuity are diretly related to machine faults that occur in production and distribution stages. In addition to economic losses, machine faults also result in industrial accidents. Early diagnosis of possible faults would cut down possible losses. To date, various solutions on fault detection has been proposed. Existing solutions either detect faults after they occur or misdiagnose them due to complexity caused by operating over multiple measurements. In this study, to the best our knowledge, we propose a supervised model that optimally determines the window size for both fault detection and classification problems. Additionally, in order to determine the features that are more heavily related with the problem, we apply the binary version (BCS) of the nature-inspired Cuckoo Search Algorithm (CSA) for feature selection. Our results indicate that determining the window size appropriately has a significant impact on accuracy and feature selection increases the F-score roughly around 13%.
  • [ X ]
    Öğe
    Monthly streamflow prediction and performance comparison of machine learning and deep learning methods
    (Springer Int Publ Ag, 2023) Ayana, Omer; Kanbak, Deniz Furkan; Keles, Muemine Kaya; Turhan, Evren
    Streamflow prediction is an important matter for the water resources management and the design of hydraulic structures that can be built on rivers. Recently, it has become a widely studied research field where data obtained from stream gauge stations can be utilized for creating estimating models by resorting to different methods such as machine and deep learning techniques. In this study, we performed monthly streamflow predictions by using the following data-driven methods of machine learning: linear regression, support vector regression, random forest and deep learning (DL) models to compare the performances of ML's and DL's techniques. A general workflow that can be applied to similar regions is presented. An estimating model containing six-input combinations and time-lagged streamflow data is improved by means of the autocorrelation function (ACF) and partial autocorrelation function (PACF). Furthermore, moving average is used as a smoothing technique to make the dataset more stable and reduce the effects of noise data. A comparative evaluation has been conducted to determine the performances of the above-mentioned methods. In this study, we proposed four different DL models and compared them with existing techniques. For the comparison of the results, we used evaluation criteria such as Nash-Sutcliffe efficiency (NSE), mean square error (MSE) and percent bias (PBIAS). The experimental results indicate that our bidirectional gated recurrent units (BiGRU) model outperforms both ML algorithms and existing solutions with 0.971 NSE, 0.001 MSE and - 1.536 PBIAS scores.