Predictive analytics, pro-active maintenance, failure prediction, performance estimation
The emergence of Internet-of-Things technologies has allowed companies to use real-time data from sensors in order to remotely monitor machines and infrastructures, to guide maintenance teams, to optimize production processes and to maximize yield. This project aimed at realizing a predictive analytics solution enabling to estimate future performance, to predict failures and to schedule pro-active maintenance based on the advanced analysis of historical data by means of simulation and modelling techniques (e.g. advanced statistics, probabilistic and machine learning models). The proposed solution was validated on energy-related production (log) data obtained from solar plants, provided by the industrial partner 3E.
The project focus on researching different algorithms and methodologies for building suitable simulation and prediction models from the data, encompassing several challenges such as : 1) the investigation of whether the use of semantic technology, such as ontologies, can provide an open and evolvable solution for the lack of harmonization and standardization in the log files within the photovoltaic domain. 2) The development of automatic methods to detect and discard irrelevant events in the log files, i.e. event parts of the regular behaviour, allowing to reduce the size of the log files and hence reduce the computation time (up to 93%) of methods leveraging these log files. 3) The combination of (heterogeneous) log files and sensor data to provide a complete and accurate performance indicator of the asset behaviour over a specified period.
January 2015 - December 2018
Projet subsidié par la Region de Bruxelles-Capital - Innoviris/project gesubsidieerd door het Brussels Hoofdstedelijk (Project subsidized by the Brussels Capital Region - Innoviris):