Amiral Technologies announces DiagFit 2.0, its blind failure prediction software. DiagFit 2.0 is developed to support industrial groups in predictive maintenance. Amiral Technologies is constantly improving its software in order to provide its customers with the best of its expertise. Thanks to customer feedback and the valuable work of its technical teams, version 2.0 of its software in Edge mode is now available.
Diagfit is a blind failure prediction software for industrial equipment based on an innovation from the CNRS. Blind mode means that the creation of the predictive models is done by learning on the healthy operating data of the equipment.
• For a given piece of equipment, Diagfit automatically generates a predictive model based on operating data in normal or “healthy data” mode.
• The data are those obtained from physical sensors in the form of “time series” that DiagFit processes with a set of powerful algorithms in order to extract any cyclicity as well as rich characteristics determining the state of health of the equipment.
Once the predictive model is built, DiagFit can monitor the equipment in “RUN” mode under user control:
• Alerts are triggered if an anomaly is detected. An indication of the type of anomaly is provided,
• Feedback from the maintenance operator on raised alerts enriches the directory of anomalies as and when,
• Prediction results are displayed, can be saved and /or exported.
These operations can be carried out in real time on data from connected equipment or in deferred mode.
DiagFit 2.0 brings the following major advances to ensure the scalability of the software as well as a user experience that is perfectly suited to the context of use:
- A complete overhaul of the user experience and graphical interface (UX / UI)
- Real-time multi-equipment, multi-sensor monitoring with different levels of monitoring (macro and granular)
- A packaged software, ready to be installed in Edge mode at customers’ premises
In addition to the consequent improvement of the architecture and user experience of DiagFit, the underlying algorithms have been refined to provide the most efficient solution on the market. Thus, a benchmark comparison of failure prediction models in unsupervised mode was carried out on a number of industrial databases. It compares a set of algorithms from the worlds of signal processing, automation, machine learning and deep learning and provides their performances against criteria that are relevant to the industry.
The results of this benchmark are published here: https://www.amiraltechnologies.com/wp2k18/wp-content/uploads/2021/05/Benchmark_AmiralTechnologies_EN-.pdf
Sebastien Le Gall, CTO of Admiral Technologies, comments: “Our teams produced a major version in close connection with the needs and feedback from our customers and partners. We are constantly developing our DiagFit software so that it best meets the demands of our customers, in terms of model performance and ergonomics of use at all levels, from maintenance operators to data science engineers. Version 2.0 allows us to have this fluidity of the interface combined with powerful features. “
DiagFit is intended for the transport, energy and manufacturing sectors. Thanks to DiagFit, customers reduce unplanned downtime, anticipate failures, understand the causes of breakdowns and potential wear.