Genetic case-based reasoning for improved mobile phone faults diagnosis

Abstract

Different types of fault diagnostic applications that utilize case-based reasoning (CBR) are applied in the diagnosis process. However, CBR cannot provide solutions to unanticipated or unknown problems. Therefore, further investigation of the retrieval and revision mechanisms of CBR is essential in improving the diagnosis accuracy and precision of the method. This study proposes a hybrid scheme that combines the genetic algorithm and CBR (GCBR) to improve CBR diagnosis. CBR applies experience and knowledge on existing cases of fault diagnosis to newly provided cases. The genetic algorithm aggregates and revises relevant cases to provide solutions to unknown cases. GCBR is implemented in a mobile phone fault diagnosis application. This domain is a good testing environment because mobile phones are of various types and models. Test results show that GCBR can detect several mobile phone faults with average accuracy 98.7%.