Design of Intelligent Fault Diagnosis System for High Lift System Based on Machine Learning

During the process of high lift system test, any improper operation during the movement of the flaps and slats, resulting in hydraulic low-pressure failure, may lead too the reduction of the braking ability or even damage of the PDU hydraulic motor, which in turn affects the dispatch rate of the aircraft. This article aims to design an intelligent fault diagnosis system for high lift system based on machine learning. The collected motor speed sensor data is transmitted to the cloud through 5G data transmission technology to form a database. After feature extraction preprocessing and training in machine learning algorithms, a fault diagnosis model is obtained, and the probability of potential faults on the PDU motor brake plates of the high lift system is calculated and predicted in real-time on cloud platform. After 100 rounds of machine learning algorithm training on the cloud platform, an average testing accuracy of 86% and an average diagnostic time of 0.7 seconds were obtained through case verification of the fault diagnosis model. This indicates that the fault diagnosis system has high diagnostic accuracy and fast speed, and can guide relevant personnel to make corresponding preparations before the fault occurs, reducing the impact on flight test and airline service. Through real-time fault diagnosis and warning, equipment that is about to malfunction can be replaced in time, reducing flight risk of the aircraft and further improving the safety and reliability of the aircraft.