Research Output
Fuzzy Logic Cascaded Current Control of DC Motor Variable Speed Drive using dSPACE
The Classification System uses a Support Vector Machine and a Decision Tree Method Based on X-Ray Images for Spinal Abnormalities
1 - 2 of 2
Publication
Two-wheel e-scooter falls under low power segment for Battery Electric Vehicle (BEV) and has gain more popularity in urban commuting. Most entry level e-scooter is still powered by DC motor due to low cost and ease of control. However basic open-loop DC Motor control employed through throttling is plugged with limited efficiency, precision, and range of speed control. Closed-loop control enables real time adjustment according to preset speed which becomes handy during auto cruising. To ensure good dynamic response, improved robustness and stable wide speed control range, a good control scheme for the motor is essential. In this project, a variable speed control scheme, namely fuzzy logic cascaded current control system was designed using MATLAB Simulink, comprising speed control loop and a current control loop 185 W Separately Excited Brushed DC Motor. The proposed control system was tested on hardware using dSPACE DS1104 platform. The system's output speed is obtained using an incremental encoder, while the output current is measured with a current sensor. Subsequently, the control system's stability, robustness, and dynamic performance were evaluated by driving the system on 120 W electrical load at varying speed. The system performance has proved superior to closed-loop by 70% on low speed ripple reduction and is on par with PI cascaded current control scheme.
Publication
A prevalent form of disorder that effects the vertebrae is a spinal disorder. X-ray technology is frequently used by medical professionals to detect abnormalities in the human body that are not visible to the unaided eye. Spinal ailment. Using the Hu moment invariant and machine learning, this study develops a system capable of feature extraction and spinal anomaly classification. The Hu Moment Invariant technique is used to derive seven moments (features) that describe an object. A support vector machine (SVM) identifies the optimal hyperplane in the input space that separates two classes. A decision tree (DT) is a technique for predicting the future by constructing a classification or regression model in the shape of a tree. Using the DT -Fine classification model derived from the Hu-Moment extraction results, the system can classify the newly developed research data in 1 minute with an accuracy of 88.2 % (highest) and 0.885782 seconds of feature extraction.