The Advanced Manufacturing Innovation Program (IMA+)  has developed Predictive Builder, a tool that uses artificial intelligence and available monitoring data to predict failures in industrial equipment.

Status: Licensed to spinoff Calmly   |   TRL 7

Problem

In many industries, the diagnosed problem is the overloading of physical assets due to meeting production targets. This overloading results in an acceleration of equipment degradation and an increase in unplanned failures, which has a negative impact on maintenance costs (representing approximately 30% to 40% of the operating cost), productivity, and safety.

Maintenance plays a vital role in addressing this problem. However, current preventive maintenance processes are based on periodic inspections and replacement of parts, which can result in replacing parts that still have a useful life. This means that failures can occur between maintenance intervals, leading to sudden and catastrophic failures.

It is crucial to anticipate failures and plan maintenance actions that avoid adverse events. Today, most equipment is being monitored online, which generates large volumes of data (Big Data). This provides a valuable opportunity to develop predictive maintenance techniques based on the Internet of Things (IoT) and Artificial Intelligence (AI). By leveraging this data and using AI algorithms, it is possible to identify patterns and early signals of potential failures, allowing proactive maintenance measures to be taken and avoiding costly interruptions in production.

Solution

The PredictiveBuilder platform is a solution that enables the implementation of machine learning models for predictive maintenance applications, leveraging the potential of the Internet of Things (IoT) and Artificial Intelligence (AI).

The platform is equipped with machine learning models specifically designed for health status monitoring on industrial equipment, have been developed as part of the Innovation in Advanced Manufacturing (IMA) Program and have been rigorously validated using data from real equipment. The results have demonstrated their superiority compared to current monitoring systems.

PredictiveBuilder consists of four fundamental modules that enable the complete process of analyzing, evaluating and detecting failures in engineering equipment. These machine learning models can be easily updated and replaced at any time from within the same application, ensuring continuous flexibility and adaptability.

Using PredictiveBuilder, maintenance teams can make informed decisions more efficiently. By anticipating potential equipment failures and degradation, maintenance actions can be proactively planned, thus avoiding unexpected and costly failures.

  • Reduce unplanned failures
  • Increase equipment safety, availability and reliability.
  • Reduce maintenance costs