Artificial intelligence to reduce gas leaks

46% more leaks detected on the network

Planning and management system of the gas network’s programmed leak research activities through machine learning algorithms aimed at optimising the effectiveness of the verifications in terms of number of leaks found

In 2019, inspections of underground and air connections continued by Hera Group’s company Inrete Distribuzione, with the introduction of a new system for the planning and management of scheduled inspections on the gas distribution network. The system consists of an artificial intelligence platform with machine learning models for the programming and optimization of inspections, completed by a web service dedicated to the operational management and reporting of the activities performed. The purpose of the system is to maximise daily inspection activities to ensure an increasing level of safety and quality of the distribution service. The activities are carried out entirely by internal staff, while planning is defined by maintenance algorithms that are dispatched to teams on cartographic support in the form of road routes optimized to minimize the routes and inspection times with a consequent increase in effectiveness and reduction the environmental impact of the research.

The data reported after the first operating year of the new gas leak management system shows the consistency between the data themselves and the aims of the programming tool. In fact, while the number of networks inspected in Emilia-Romagna in 2019 was about 4% lower than the previous year, the number of reported losses increased by about 46% compared to 2018. In addition to pursuing industrial efficiency objectives, this performance also aims at a constant improvement in company safety standards, which can be deduced from the decrease of about 18% of the dispersions reported by third parties.

Area: province of Bologna, Ferrara, Forlì-Cesena, Modena, Ravenna, Rimini