Welcome to i4QLRT

The aim of the described algorithm is to develop a predictive model to anticipate failures in computer numerical control (CNC) machines. This model is intended to support the transition from a corrective maintenance to a preventive strategy, alerting operators about possible failures before they occur. In this way, it aims to improve process quality, reduce downtime and minimise defective parts.

The algorithm uses signal and alarm data from CNC machines to learn to identify situations that can lead to failures. As soon as it detects a potential anomaly, the model notifies operators in real time, allowing them to autonomously perform maintenance tasks and make informed decisions to prevent equipment failure. This not only improves process efficiency, but also empowers operators with real-time data and predictive insights.

The predictive model is based on long and short-term memory (LSTM) recurrent neural networks. These networks are especially suited to handle sequential data and capture long-term dependencies, which is crucial for analysing signals from multiple sensors on CNC machines.