Artificial neural networks function on the principle of artificial intelligence that mimics human work. The first idea of neural networks appeared in the mid-19th century, and today it is of great importance in diagnosing faults in the ship's system. Faults are present in all systems, and they are characterized by the termination of the system or interference with its normal operation. A neural network-based diagnostic system detects damage, malfunction, or downtime based on the principle of the data it processes, and displays it as information. There are different methods and strategies of system maintenance, and the most common maintenance methods are: corrective, preventive and planned maintenance. The conduct of the diagnostic procedure is based on the beginning of data processing, extraction of important data, assessment of the working condition and prediction of the remaining service life without failures. A significant achievement in the maritime system are autonomous ships operating on the principle of artificial intelligence. It is characteristic of this type of ship that they already have a built-in diagnostic system, and that their maintenance is significantly reduced than is the case with other ships.