AUDEI | Automated detection of energy efficiency deficits at industrial sites
In industry, natural ageing effects cause energy efficiency deficits that often go unnoticed. The aim of the project is to detect such efficiency deficits automatically by using machine learning methods.
01.10.2017 - 30.09.2019
- Increasing energy efficiency is a central task for the future.
- However, these efforts are being counteracted by external influences:
- creeping occurrence: natural wear and tear and maintenance problems (leaks in compressed air networks, clogged filters)
- sudden occurrence: lack of control over the setting of equipment (reconfiguration for special use or maintenance without resetting to previous status)
- For the automatic detection of energy efficiency deficits, a prototypical development and testing of software services based on anomaly detection methods will be carried out.
- In contrast to the software solutions offered on the market, no static and manual settings of limit values for measurement series are necessary.
- Selection and manual analysis of data sets
- Prioritization of relevant errors
- Development of anomaly detection methods
- Application and evaluation of developed methods
- Development of a software prototype
- Avoiding CO2 emissions through more efficient operations
- Annual energy savings of approx. 5 %: calculated with 300 customers, this corresponds to a saving of the annual electricity requirements of a small town (approx. 14,000 private households)
- deZem GmbH
- Siemens Real Estate
- Sika Deutschland GmbH
- Bayer AG