Fault Detection and Diagnosis in Automotive Control Systems using Machine Learning

The Software Fault detection and diagnosis approaches supports to find the fault vulnerable constituents in the software development in early stages. An effective diagnosis approach can support test administrators to locate defects and defect-vulnerable software modules. The Feature Extraction (FE) process is the applicable solution to solve high dimensionality and polynomial time complexity issue in the prediction of software defect. In this research, a new fault detection approach is proposed by utilizing the Stacked Auto Encoder (SAE) with Support Vector Machine (SVM) and Artificial Neural Network (ANN). In this approach, the ANN is utilized for the diagnosis process to distribute the data named the fault type. Now, the input generates the Deep Neural Network as well as fuzzy depiction process. the outcome of these approaches is then integrated by the dense connected layers. The proposed SAE based SVM-ANN model attains better results by utilizing evaluation metrics like Accuracy, Precision, F1-score, and Area Under Curve (AUC) values about 99.87%, 99.52%, 99.82% and 0.99 respectively which is comparatively higher than existing techniques like Unsupervised Learning, SVM, Generative Adversarial Neural Network.