<![CDATA[In the digital era, the need for technology-based monitoring systems is increasing, especially in the transportation and logistics industry. PT BDR faces challenges in reporting and monitoring fuel consumption for operational vehicles due to the manual recording process, making it difficult to detect anomalous data and making monitoring less efficient. This study aims to develop a web-based application capable of automatically monitoring and detecting fuel consumption anomalies by utilizing machine learning and deep learning technologies. The system development method uses the CRISP-DM approach, which includes the stages of business understanding, data understanding, data preparation, modeling, evaluation, and implementation. The Isolation Forest algorithm is used to detect anomalies based on fuel volume data, mileage, and vehicle consumption ratio, while the MobileNetV2-based Content-Based Image Retrieval (CBIR) method is applied to validate the suitability of gas station photos. The trained model is then integrated into the API using the Flask framework, with testing conducted through blackbox and whitebox testing. The test results show that the system is able to detect anomalies with a good level of accuracy and can be used practically by users. The implementation of this application is expected to improve the company's operational efficiency, reduce potential losses due to fuel misappropriation, and support the digitalization of the fuel monitoring process to be more accurate, effective, and integrated.]]>
Copyrights © 2025
