Repository logo
 

Faculty of Engineering and Built Environment

Permanent URI for this communityhttp://ir-dev.dut.ac.za/handle/10321/9

Browse

Search Results

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Autonomous switching of electric locomotives in neutral sections
    (2023-05) Mcineka, Christopher Thembinkosi; Pillay, Nelendran; Reddy, Serendra
    Abstract Electrical locomotives traversing in a neutral section must switch off as they enter a different phase voltage. The current system used to auto-switch these electric locomotives requires two pairs of induction magnets installed adjacent in-between the rails and two sensors installed underneath the locomotives. However, the return cost of investment is low, maintenance costs increase due to failures, and locomotives do not auto-switch due to the degradation of magnet strength. Additionally, damage to sensors due to animal collisions or objects also causes switching failures, and vandalism and theft are some of the challenges limiting this switching scheme. Furthermore, the latter switching method does not align with the Transnet 4.0 strategy aimed at adopting the Fourth Industrial Revolution (41R). Therefore, to align with the Fourth Industrial Revolution, this research proposed a computer vision-based approach to switch electric locomotives automatically. The requirements are a computer, a high-definition camera, and open and close markers. While the latter gives an overview of the hardware used, creating a new dataset with training and testing images allowed for developing a machine learning classification model. Firstly, image pre-processing converts the RGB images to greyscale then the noise is removed using a bilateral filter. Secondly, segmentation and marker extraction is performed by employing the Sobel operator and Circular Hough Transform. Thirdly, features are extracted using a Histogram of Oriented Gradients and employing Linear Support Vector Machine to perform classification. However, before selecting the latter classifier, the feature extractor is tested against Quadratic Support Vector Machine, K-Nearest Neighbour and Convolutional Neural Network. The model's accuracy is then measured using the training set and ground truth dataset. The test set is used to validate the model with evaluation methods such as a confusion matrix, Fl-measure and 2-fold cross­ validation.
  • Thumbnail Image
    Item
    Machine learning classifiers based on HoG features extracted from locomotive neutral section images
    (IEEE, 2022-10-27) Mcineka, Christopher Thembinkosi; Pillay, Nelendran
    This paper presents a comparative study on machine learning algorithms for neutral section image classification. The classifiers are trained by employing the Histogram of Oriented Gradient features that are extracted from the neutral section dataset [1]. A neutral section is a phase break that is used on the Transnet freight rail system to separate the single-phase supply from the 25kV three-phase overhead traction supply. The 25kV is a stepped-down voltage from an 88kV three-phase supply coming from the national grid. While the main purpose of the neutral section is to separate phase voltages, electric locomotives can traverse through these phases by switching On and Off. This auto-switching is possible through induction magnets installed in between the rails and with magnet detection sensors installed underneath the locomotives. However, a computer vision model has been developed, trained, and tested with a neutral section dataset containing images having open and close markers [1]. This paper, therefore, utilises this dataset to provide performance comparison on several machine learning classification algorithms viz. Decision Tree, Discriminant Analysis, Support Vector Machine, K-Nearest Neighbors, Ensemble, Naïve Bayes, and Convolutional Neural Network. A confusion matrix, F1- measure and computation time are employed to measure the performance of each classifier. The MATLAB Classification Learner application was used to obtain the results. The results show that the Linear Support Vector Machine performs best when considering performance and prediction speed. The Linear Support Vector Machine achieved a training accuracy of 93.40% with a test accuracy reaching 94% at a prediction speed of 75 objects per second (computation time).