Gompertz binary particle swarm optimization and support vector data description system for fault detection and feature selection applied in automotive pedals components

This work presents an improved fault detection by reference space optimization and simultaneous feature selection applied in a manufacturing complex process of automotive pedals components. Support vector data description (SVDD) one-class classification method uses a hypersphere with the minimum volume to find an enclosed boundary containing almost all target objects. Gompertz binary particle swarm optimization algorithm (GBPSO) is applied to optimize kernel hyperparameters for SVDD and simultaneously solve the feature selection problem. In order to justify and validate the results, also the genetic algorithm (GA) and binary particle swarm optimization algorithm (BPSO) are presented to compare the performances of the three approaches in terms of the misclassification function. The experimental results showed that the proposed approach can correctly select the influencing input variables in order to achieve an efficient fault detection.