Predictive maintenance in the industry
Predictive maintenance “is the asset management practice of repairing an asset or piece of equipment before it fails based on data received about it” (source, IBM).
This practice aims to rationalize costs, maintain operational stability and increase profitability by limiting unwanted stoppages.
Many companies offer predictive maintenance solutions for industry and position themselves on the challenges of the industry of the future. Among them, Amiral Technologies specializes in the failure prediction for industrial equipment.
An observation is made by the entire industry:
- there is a wide variety of industrial equipment and sensors
- the data generated by the sensors measuring the physical quantities of the equipment are mainly industrial time series, by nature complex to analyze
- and finally there are few or no occurrences of failures in the historical data, thus making the predictive models long to deploy
The only approach to meet these three constraints, is Amiral Technologies’ DiagFit software. It consists of implementing a failure prediction software solution based on the creation of a normality space specific to the equipment, created from healthy data from its sensors. It is a process commonly called “anomaly detection” in machine learning which is based on so-called unsupervised artificial intelligence algorithms.
Anomaly detection with DiagFit
Blind mode means that Amiral Technologies’ DiagFit software does not need to know the type of equipment it is monitoring, and does not need historical fault data to produce predictive models. This operation allows extremely fast implementation of the software.
Thanks to its “unsupervised” approach, the software learns on the healthy data of the equipment and automatically builds a robust and reliable space of normality (multidimensional space where the operation of a healthy equipment is well defined). This operation allows the tool to detect any deviation from normality, which will be called “anomaly”. Thanks to the inventions – resulting from scientific research – exploited by the software, the detection of anomalies is carried out with high precision and a low rate of false alarms.
The objective is that this construction is done without code manipulation by the domaine expert user, fast, precise, agnostic and above all interpretable by users.
Blind mode, a unique innovation
Blind failure prediction is therefore the ability to go from anomaly detection without historical failure data (the first blind level), and without knowing either the types of sensors or the types of equipment (the second level of blind), to the prediction of failures never encountered before thanks to other mechanisms available to the user when exploiting the normality space in operational mode.
Finally, the user retains throughout the process the ability to interpret the anomalies detected, and therefore controls the decision-making process of the technology.
