<![CDATA[Motorcycles represent the dominant mode of transportation in Indonesia, yet user comprehension of vehicle technical conditions remains inadequate. Limited access to competent mechanical expertise delays fault identification and increases the risk of accelerating damage. This study develops a desktop-based expert system named Apparatus Cura capable of automatically diagnosing engine failures in automatic and manual (bebek) motorcycle types. The system integrates three inference methods: Forward chaining for symptom-to-fault tracing, backward chaining for fault hypothesis verification, and Certainty factor (CF) for quantifying diagnostic confidence. The knowledge base was constructed from 25 symptoms and 26 fault types for automatic motorcycles, and 43 symptoms and 35 fault types for manual motorcycles, formulated in collaboration with experienced mechanics in the form of more than 60 IF-THEN rules. The user interface was developed using the CustomTkinter library in Python. Functional testing across 8 scenarios yielded a 100% success rate. Accuracy evaluation on 30 test cases produced a diagnostic correctness rate of 83.3%. The system successfully integrates all three inference methods within a unified platform, providing probability estimates that closely approximate assessments made by professional mechanics.]]>
Copyrights © 2026
