HERAtech, the technological heart of the HERA Group, needed to:
Dinova implemented a machine learning solution and a streaming analytics platform tailored for HERAtech, achieving:
In the water mains systems, fluctuations in chlorine levels often exceeded preset ranges, generating false alarms. This required hours of work by operations teams, required to verify every anomaly.
HERAtech si è rivolta a Dinova per ridurre gli allarmi non pertinenti e ottimizzare le risorse, migliorando il monitoraggio in tempo reale e aumentando la capacità predittiva di eventuali guasti.
The project for HERAtech was carried out in two phases: after the initial development of a Proof of Concept (POC), an ad hoc tool was implemented, designed according to the specific configuration of the plants and aimed at monitoring the chlorine injection parameter in the water.
The integration of Machine Learning into the solution allows HERAtech to identify false alarms through an algorithm, trained using supervised learning, which has learned all alarm situations, distinguishing those that required operator intervention from those in which the alert was not relevant.
To support the algorithm, Dinova developed a streaming analytics platform capable of acquiring the historical data from the sensors, automatically and in real time analyzing the hundreds of values collected across the entire area and reporting whether a particular alarm requires intervention.
Subsequently, a new front-end interface was also created to simplify the monitoring of chlorine levels and easily share data with operators, maintenance personnel, and analysis laboratories. The user-centered interface was designed to provide each individual operator with a transparent overview of the analysis status and, for this reason as well, was successfully integrated with SCADA, the currently active alarm management system.
The adoption of Machine Learning technology has had positive repercussions on the business as well: the extensive collection and processing of data through the ML algorithm has in fact led to an increase in the predictive capacity of the alarm systems, allowing savings in terms of emergency management costs.
"The number of false positives has been reduced by 85%, with a rising trend. This has allowed operators to focus more on other activities, such as answering phone calls: on average, each false alarm took about four minutes to handle, not to mention that in at least 15% of cases, it was not possible to remotely assess the alert and an on-site intervention was required to verify the nature of the alarm. Furthermore, the introduction of the new tool has led to an increase in the predictive capacity for potential faults. The platform is also capable of processing and returning a large amount of valuable data in real time, proving to be an excellent support for decision-making processes."
— Sandro Boarini, dirigente responsabile Polo Telecontrollo e Call center Tecnico Hera
Innovation and artificial intelligence for efficient and predictive management
Dinova was the ideal choice for HERAtech thanks to its ability to integrate advanced machine learning technologies with customized and scalable solutions. The streaming analytics platform and user-centered interface made chlorine level monitoring more effective and intuitive, dramatically reducing false alarms and optimizing company resources.
The adoption of the solution has brought significant benefits:
This project, awarded the Smau Innovation Award and the Digital360 Awards 2020, demonstrates how the application of Machine Learning and analytics can enhance business productivity, optimize operational costs, and promote sustainable innovation in public services.
Connect with stories, visions, and tools to truly innovate.
Dinova Srl – Company subject to the management and coordination of Maggioli S.p.A.
C.F. e P. Iva: 03008670360 – Registered and operational office: Via dei Lapidari 12 – 40129 Bologna, Italy