Exploring the Application of CBM+ in an ISR Weapon System, Thermal Observation Device

Abstract

Purpose: This study suggests a method to apply the condition-based maintenance plus (CBM+) concept, known to improve system availability, by introducing a deep learning model on the thermal observation device (TOD)—representative of an intelligence, surveillance, and reconnaissance (ISR) weapon system.BRMethods: The temperature was chosen as a health index, and the data (normal/abnormal) was acquired by operating the device in a controlled laboratory and from the manufacturer.BRResults: The criteria to classify the abnormal condition were established regarding time and temperature. The following deep learning models were introduced: the convolution neural network (CNN) model for detecting the abnormality and long short-term memory (LSTM) for predicting remaining useful life (RUL). The models proved that the accuracy and loss were fine. Thus, the logic flow of CBM+ for the TOD was suggested by incorporating the models.BRConclusion: The data collected had limitations due to the highly controlled data acquisition. Nevertheless, the possibility of applying the CBM+ logic to similar weapon systems that employ the cooled-type detector could be confirmed.