CaVision

The goal of the CaVision project is to develop new independent maintenance services based on an example of a cavitation detection module using a Machine Learning approach in order to participate in the implementation of predictive maintenance in hydroelectric power plants and thus increase their flexibility.

Abstract

With the massive integration of new renewable energies, the flexibility of hydroelectric power plants is crucial to maintain the stability of the network. To increase this flexibility, hydraulic machines are used outside their optimal operating range where complex physical phenomena appear. Among them, cavitation which can cause high wear and tear and vibrations and requires regular and costly overhauls. In order to complete the monitoring of machines that will be increasingly subject to this type of wear, HYDRO Exploitation SA, responsible for the operation and maintenance of 22% of the installed capacity of the Swiss hydroelectric park, approached the HES SO VS to develop a cavitation monitoring solution to complement their monitoring tools and provide a service to validate the flexible operation of a power plant. A low-cost and easily transportable sound cavitation detection module will be developed. The frequency spectrum of the sound signal will be processed by an innovative adaptation of statistical analysis methods, allowing the identification and visualization of different cavitation regimes. The analysis data will be used to precisely identify the operating points responsible for defects, potential failures and premature wear in order to facilitate predictive maintenance. The use of Machine Learning and Deep Learning algorithms will then be used to build predictive models from the data acquired during the use of this new module over the long term.

Damage caused by cavitation

Partners

•  HYDRO Exploitation SA

Financing

Publications

• 2020 Master HES-SO Engineering - D. Voide - Machine Learning pour détecter la cavitation
• 2019 Bachelor - F. Pedroni Détection de la cavitation

Photo: SNSF, Scientific Image Contest 2019, Cavitating Flow over a Hydrofoil by Ali Amini, EPFL