SYSTEM FOR PREDICTING ENGINE FAILURE USING BIG DATA

摘要

The present invention relates to an engine equipment failure prediction system using big data, which collects sensor data from a sensor module installed in a ship engine and converts and normalizes the collected data into a form useful for failure prediction. And a correlation analysis unit that extracts important factors affecting engine data through correlation analysis among the data collected and preprocessed through the preprocessing unit, the data collection and preprocessing unit, and an encoder-decoder method with the extracted important factors. A fault diagnosis model generator that extracts a previously unknown fault pattern and generates a fault diagnosis model of the engine, and a fault diagnosis that determines faults and outliers by comparing the reconstruction error in the encoder-decoder method with a preset threshold value. By including wealth, it enables general-purpose and data-adaptive analysis, and even if there are no outliers or rare, it is possible to perform anomaly detection through self-learning.