As the performance of aviation equipment continues to improve, it brings serious challenges and burdens to testing, diagnosis, and maintenance. Fault prediction technology, as a key support for the health management of aviation equipment, can predict the health status and potential failures of devices in advance, effectively reducing the risk of faults, conserving resources, and minimizing economic losses. Leveraging the efficiency of deep learning in knowledge discovery, a fault prediction method for aviation equipment based on knowledge discovery is proposed. This method uses deep learning to mine fault-related knowledge and establish a fault prediction model. Through the construction, optimization, and continuous evolution of rule-based knowledge and model-based knowledge, the inference engine provides predictive support for different faults and their uncertain factors, thereby offering strong guarantees for aircraft performance, health status, and safety assurance.